Model save

README.md

@@ -27,17 +27,17 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/jinson/huggingface/runs/qrc2c3at) 
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/jinson/huggingface/runs/5ajy3q44) 
 
 
 This model was trained with SFT.
 
 ### Framework versions
 
-- TRL: 0.16.0.dev0
-- Transformers: 4.49.0.dev0
+- TRL: 0.15.0
+- Transformers: 4.50.0.dev0
 - Pytorch: 2.5.1
-- Datasets: 3.3.0
+- Datasets: 3.3.2
 - Tokenizers: 0.21.0
 
 ## Citations

all_results.json

@@ -1,8 +1,8 @@
 {
-    "total_flos": 76560728064000.0,
-    "train_loss": 0.8027291595935822,
-    "train_runtime": 10618.3737,
-    "train_samples": 16610,
-    "train_samples_per_second": 2.036,
-    "train_steps_per_second": 0.016
+    "total_flos": 488165445992448.0,
+    "train_loss": 0.5792719663968727,
+    "train_runtime": 25010.1039,
+    "train_samples": 93733,
+    "train_samples_per_second": 1.371,
+    "train_steps_per_second": 0.011
 }

generation_config.json

@@ -10,5 +10,5 @@
   "temperature": 0.7,
   "top_k": 20,
   "top_p": 0.8,
-  "transformers_version": "4.49.0.dev0"
+  "transformers_version": "4.50.0.dev0"
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "total_flos": 76560728064000.0,
-    "train_loss": 0.8027291595935822,
-    "train_runtime": 10618.3737,
-    "train_samples": 16610,
-    "train_samples_per_second": 2.036,
-    "train_steps_per_second": 0.016
+    "total_flos": 488165445992448.0,
+    "train_loss": 0.5792719663968727,
+    "train_runtime": 25010.1039,
+    "train_samples": 93733,
+    "train_samples_per_second": 1.371,
+    "train_steps_per_second": 0.011
 }

trainer_state.json

@@ -1,316 +1,413 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9944506104328524,
-  "eval_steps": 100,
-  "global_step": 168,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 268,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.029596744358120607,
-      "grad_norm": 2.297286039806607,
-      "learning_rate": 5.882352941176471e-06,
-      "loss": 1.1002,
-      "mean_token_accuracy": 0.7101261564901226,
+      "epoch": 0.018656716417910446,
+      "grad_norm": 1.8357200745044757,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.8456,
       "step": 5
     },
     {
-      "epoch": 0.059193488716241215,
-      "grad_norm": 1.598959318938298,
-      "learning_rate": 1.1764705882352942e-05,
-      "loss": 1.0325,
-      "mean_token_accuracy": 0.7202338207680274,
+      "epoch": 0.03731343283582089,
+      "grad_norm": 0.7880079087520768,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.7682,
       "step": 10
     },
     {
-      "epoch": 0.08879023307436182,
-      "grad_norm": 0.8557338569703955,
-      "learning_rate": 1.7647058823529414e-05,
-      "loss": 0.9517,
-      "mean_token_accuracy": 0.7325525081674223,
+      "epoch": 0.055970149253731345,
+      "grad_norm": 0.5072079736182407,
+      "learning_rate": 4.999827900623038e-05,
+      "loss": 0.7026,
       "step": 15
     },
     {
-      "epoch": 0.11838697743248243,
-      "grad_norm": 0.6524948891488312,
-      "learning_rate": 1.9980527694749952e-05,
-      "loss": 0.8805,
-      "mean_token_accuracy": 0.7473646233841255,
+      "epoch": 0.07462686567164178,
+      "grad_norm": 0.3956980910828853,
+      "learning_rate": 4.993807186343243e-05,
+      "loss": 0.6745,
       "step": 20
     },
     {
-      "epoch": 0.14798372179060304,
-      "grad_norm": 0.5586962091196656,
-      "learning_rate": 1.986180478852149e-05,
-      "loss": 0.8535,
-      "mean_token_accuracy": 0.7518094119293168,
+      "epoch": 0.09328358208955224,
+      "grad_norm": 0.38447381525886887,
+      "learning_rate": 4.979207812402531e-05,
+      "loss": 0.6436,
       "step": 25
     },
     {
-      "epoch": 0.17758046614872364,
-      "grad_norm": 0.4501992507045545,
-      "learning_rate": 1.963645895935632e-05,
-      "loss": 0.8403,
-      "mean_token_accuracy": 0.7545180402865705,
+      "epoch": 0.11194029850746269,
+      "grad_norm": 0.299192319561256,
+      "learning_rate": 4.956085596012407e-05,
+      "loss": 0.6362,
       "step": 30
     },
     {
-      "epoch": 0.20717721050684423,
-      "grad_norm": 0.40514440229124343,
-      "learning_rate": 1.930692657985482e-05,
-      "loss": 0.8211,
-      "mean_token_accuracy": 0.7579922643362512,
+      "epoch": 0.13059701492537312,
+      "grad_norm": 0.2848642692822099,
+      "learning_rate": 4.924528939432311e-05,
+      "loss": 0.6199,
       "step": 35
     },
     {
-      "epoch": 0.23677395486496486,
-      "grad_norm": 0.39471525110138866,
-      "learning_rate": 1.887677045685188e-05,
-      "loss": 0.8049,
-      "mean_token_accuracy": 0.7621190554838779,
+      "epoch": 0.14925373134328357,
+      "grad_norm": 0.246624944967583,
+      "learning_rate": 4.884658491984735e-05,
+      "loss": 0.6105,
       "step": 40
     },
     {
-      "epoch": 0.2663706992230855,
-      "grad_norm": 0.43739042181311494,
-      "learning_rate": 1.8350641311400813e-05,
-      "loss": 0.7984,
-      "mean_token_accuracy": 0.7634295080529785,
+      "epoch": 0.16791044776119404,
+      "grad_norm": 0.2113047031719804,
+      "learning_rate": 4.8366266887814235e-05,
+      "loss": 0.6112,
       "step": 45
     },
     {
-      "epoch": 0.2959674435812061,
-      "grad_norm": 0.4050015485090704,
-      "learning_rate": 1.773422749654988e-05,
-      "loss": 0.7936,
-      "mean_token_accuracy": 0.7642335004344475,
+      "epoch": 0.1865671641791045,
+      "grad_norm": 0.24707697994991026,
+      "learning_rate": 4.780617167924209e-05,
+      "loss": 0.5938,
       "step": 50
     },
     {
-      "epoch": 0.3255641879393267,
-      "grad_norm": 0.3862008743204375,
-      "learning_rate": 1.7034193496547903e-05,
-      "loss": 0.7894,
-      "mean_token_accuracy": 0.7648157866255213,
+      "epoch": 0.20522388059701493,
+      "grad_norm": 0.2596596831883685,
+      "learning_rate": 4.716844068408693e-05,
+      "loss": 0.5965,
       "step": 55
     },
     {
-      "epoch": 0.3551609322974473,
-      "grad_norm": 0.3808360474271673,
-      "learning_rate": 1.6258107872407376e-05,
-      "loss": 0.7999,
-      "mean_token_accuracy": 0.7619282901641846,
+      "epoch": 0.22388059701492538,
+      "grad_norm": 0.26985436915379585,
+      "learning_rate": 4.6455512114150546e-05,
+      "loss": 0.5918,
       "step": 60
     },
     {
-      "epoch": 0.38475767665556787,
-      "grad_norm": 0.41625214344673966,
-      "learning_rate": 1.5414361432856475e-05,
-      "loss": 0.7869,
-      "mean_token_accuracy": 0.7646248305016428,
+      "epoch": 0.24253731343283583,
+      "grad_norm": 0.5749756625596264,
+      "learning_rate": 4.5670111681161296e-05,
+      "loss": 0.583,
       "step": 65
     },
     {
-      "epoch": 0.41435442101368847,
-      "grad_norm": 0.3836610280759421,
-      "learning_rate": 1.4512076515391375e-05,
-      "loss": 0.7782,
-      "mean_token_accuracy": 0.7674149625106065,
+      "epoch": 0.26119402985074625,
+      "grad_norm": 0.4737380183827698,
+      "learning_rate": 4.481524217566783e-05,
+      "loss": 0.58,
       "step": 70
     },
     {
-      "epoch": 0.4439511653718091,
-      "grad_norm": 0.3429055885862389,
-      "learning_rate": 1.356100835825547e-05,
-      "loss": 0.7771,
-      "mean_token_accuracy": 0.767463215138475,
+      "epoch": 0.2798507462686567,
+      "grad_norm": 0.4262259418885374,
+      "learning_rate": 4.3894171986588217e-05,
+      "loss": 0.5785,
       "step": 75
     },
     {
-      "epoch": 0.4735479097299297,
-      "grad_norm": 0.3548520942127707,
-      "learning_rate": 1.257143962968246e-05,
-      "loss": 0.7783,
-      "mean_token_accuracy": 0.7665814820650783,
+      "epoch": 0.29850746268656714,
+      "grad_norm": 0.43257212906351034,
+      "learning_rate": 4.29104226053073e-05,
+      "loss": 0.5775,
       "step": 80
     },
     {
-      "epoch": 0.5031446540880503,
-      "grad_norm": 0.34670016990380903,
-      "learning_rate": 1.155406925472205e-05,
-      "loss": 0.7786,
-      "mean_token_accuracy": 0.7668829515693034,
+      "epoch": 0.31716417910447764,
+      "grad_norm": 0.32488482534238494,
+      "learning_rate": 4.186775516209732e-05,
+      "loss": 0.5728,
       "step": 85
     },
     {
-      "epoch": 0.532741398446171,
-      "grad_norm": 0.354375470838285,
-      "learning_rate": 1.0519896741619803e-05,
-      "loss": 0.7724,
-      "mean_token_accuracy": 0.7681869031104136,
+      "epoch": 0.3358208955223881,
+      "grad_norm": 0.3511807271347856,
+      "learning_rate": 4.077015604633669e-05,
+      "loss": 0.5752,
       "step": 90
     },
     {
-      "epoch": 0.5623381428042915,
-      "grad_norm": 0.35042728553773317,
-      "learning_rate": 9.480103258380198e-06,
-      "loss": 0.7668,
-      "mean_token_accuracy": 0.7697937295203819,
+      "epoch": 0.35447761194029853,
+      "grad_norm": 0.31519292419205,
+      "learning_rate": 3.962182166550441e-05,
+      "loss": 0.5737,
       "step": 95
     },
     {
-      "epoch": 0.5919348871624122,
-      "grad_norm": 0.3487619390880057,
-      "learning_rate": 8.445930745277953e-06,
-      "loss": 0.7584,
-      "mean_token_accuracy": 0.7713990834609527,
+      "epoch": 0.373134328358209,
+      "grad_norm": 0.27842672077069447,
+      "learning_rate": 3.8427142401220634e-05,
+      "loss": 0.5697,
       "step": 100
     },
     {
-      "epoch": 0.5919348871624122,
-      "eval_loss": 0.7887204885482788,
-      "eval_mean_token_accuracy": 0.7581970585874204,
-      "eval_runtime": 20.8083,
-      "eval_samples_per_second": 6.199,
-      "eval_steps_per_second": 0.817,
-      "step": 100
-    },
-    {
-      "epoch": 0.6215316315205327,
-      "grad_norm": 0.3648322197104147,
-      "learning_rate": 7.428560370317542e-06,
-      "loss": 0.7592,
-      "mean_token_accuracy": 0.77141248230712,
+      "epoch": 0.3917910447761194,
+      "grad_norm": 0.258137938167996,
+      "learning_rate": 3.71906858236735e-05,
+      "loss": 0.566,
       "step": 105
     },
     {
-      "epoch": 0.6511283758786534,
-      "grad_norm": 0.3244685784676015,
-      "learning_rate": 6.438991641744531e-06,
-      "loss": 0.7777,
-      "mean_token_accuracy": 0.7663982071721088,
+      "epoch": 0.41044776119402987,
+      "grad_norm": 0.24782019982832532,
+      "learning_rate": 3.591717922860785e-05,
+      "loss": 0.5732,
       "step": 110
     },
     {
-      "epoch": 0.680725120236774,
-      "grad_norm": 0.31476170427058586,
-      "learning_rate": 5.487923484608629e-06,
-      "loss": 0.7637,
-      "mean_token_accuracy": 0.7702532985281345,
+      "epoch": 0.4291044776119403,
+      "grad_norm": 0.21367211036985273,
+      "learning_rate": 3.46114915636416e-05,
+      "loss": 0.5641,
       "step": 115
     },
     {
-      "epoch": 0.7103218645948945,
-      "grad_norm": 0.30378046295629435,
-      "learning_rate": 4.5856385671435285e-06,
-      "loss": 0.7512,
-      "mean_token_accuracy": 0.7742151065835647,
+      "epoch": 0.44776119402985076,
+      "grad_norm": 0.2359860351856442,
+      "learning_rate": 3.3278614813010034e-05,
+      "loss": 0.565,
       "step": 120
     },
     {
-      "epoch": 0.7399186089530152,
-      "grad_norm": 0.3245568919497631,
-      "learning_rate": 3.7418921275926245e-06,
-      "loss": 0.7596,
-      "mean_token_accuracy": 0.7716760745454179,
+      "epoch": 0.4664179104477612,
+      "grad_norm": 0.23762377488074504,
+      "learning_rate": 3.1923644911909e-05,
+      "loss": 0.5619,
       "step": 125
     },
     {
-      "epoch": 0.7695153533111357,
-      "grad_norm": 0.32838254037360903,
-      "learning_rate": 2.965806503452098e-06,
-      "loss": 0.7626,
-      "mean_token_accuracy": 0.7704949384884608,
+      "epoch": 0.48507462686567165,
+      "grad_norm": 0.21798417043326976,
+      "learning_rate": 3.0551762263406576e-05,
+      "loss": 0.5606,
       "step": 130
     },
     {
-      "epoch": 0.7991120976692564,
-      "grad_norm": 0.30665122068538603,
-      "learning_rate": 2.265772503450122e-06,
-      "loss": 0.7686,
-      "mean_token_accuracy": 0.7688054766465177,
+      "epoch": 0.503731343283582,
+      "grad_norm": 0.2185282511977544,
+      "learning_rate": 2.9168211932412042e-05,
+      "loss": 0.5578,
       "step": 135
     },
     {
-      "epoch": 0.8287088420273769,
-      "grad_norm": 0.2993142489345467,
-      "learning_rate": 1.6493586885991908e-06,
-      "loss": 0.7599,
-      "mean_token_accuracy": 0.771630464199698,
+      "epoch": 0.5223880597014925,
+      "grad_norm": 0.2017005705140969,
+      "learning_rate": 2.777828359242567e-05,
+      "loss": 0.5632,
       "step": 140
     },
     {
-      "epoch": 0.8583055863854976,
-      "grad_norm": 0.31657760951455227,
-      "learning_rate": 1.1232295431481222e-06,
-      "loss": 0.7598,
-      "mean_token_accuracy": 0.7713579193104667,
+      "epoch": 0.5410447761194029,
+      "grad_norm": 0.23403016673929508,
+      "learning_rate": 2.6387291301738377e-05,
+      "loss": 0.5559,
       "step": 145
     },
     {
-      "epoch": 0.8879023307436182,
-      "grad_norm": 0.29677317819620136,
-      "learning_rate": 6.930734201451817e-07,
-      "loss": 0.7508,
-      "mean_token_accuracy": 0.7738701734792748,
+      "epoch": 0.5597014925373134,
+      "grad_norm": 0.20972754184297304,
+      "learning_rate": 2.50005531864019e-05,
+      "loss": 0.5537,
       "step": 150
     },
     {
-      "epoch": 0.9174990751017388,
-      "grad_norm": 0.28445809248236675,
-      "learning_rate": 3.635410406436857e-07,
-      "loss": 0.7413,
-      "mean_token_accuracy": 0.7766777954365708,
+      "epoch": 0.5783582089552238,
+      "grad_norm": 0.18963422103493383,
+      "learning_rate": 2.362337110764688e-05,
+      "loss": 0.554,
       "step": 155
     },
     {
-      "epoch": 0.9470958194598594,
-      "grad_norm": 0.30204735729222293,
-      "learning_rate": 1.3819521147851122e-07,
-      "loss": 0.7543,
-      "mean_token_accuracy": 0.7731135856147668,
+      "epoch": 0.5970149253731343,
+      "grad_norm": 0.19986620919481152,
+      "learning_rate": 2.226101039148557e-05,
+      "loss": 0.5522,
       "step": 160
     },
     {
-      "epoch": 0.97669256381798,
-      "grad_norm": 0.2818907779460126,
-      "learning_rate": 1.947230525005006e-08,
-      "loss": 0.747,
-      "mean_token_accuracy": 0.7748765676578209,
+      "epoch": 0.6156716417910447,
+      "grad_norm": 0.22264982729446323,
+      "learning_rate": 2.0918679697998252e-05,
+      "loss": 0.5511,
       "step": 165
     },
     {
-      "epoch": 0.9944506104328524,
-      "mean_token_accuracy": 0.7702185157826186,
-      "step": 168,
-      "total_flos": 76560728064000.0,
-      "train_loss": 0.8027291595935822,
-      "train_runtime": 10618.3737,
-      "train_samples_per_second": 2.036,
-      "train_steps_per_second": 0.016
+      "epoch": 0.6343283582089553,
+      "grad_norm": 0.2080998889583045,
+      "learning_rate": 1.9601511107268255e-05,
+      "loss": 0.5515,
+      "step": 170
+    },
+    {
+      "epoch": 0.6529850746268657,
+      "grad_norm": 0.19343233196548346,
+      "learning_rate": 1.8314540498102216e-05,
+      "loss": 0.5512,
+      "step": 175
+    },
+    {
+      "epoch": 0.6716417910447762,
+      "grad_norm": 0.17540535388811454,
+      "learning_rate": 1.7062688294552992e-05,
+      "loss": 0.5433,
+      "step": 180
+    },
+    {
+      "epoch": 0.6902985074626866,
+      "grad_norm": 0.19756929408176382,
+      "learning_rate": 1.5850740653856096e-05,
+      "loss": 0.5467,
+      "step": 185
+    },
+    {
+      "epoch": 0.7089552238805971,
+      "grad_norm": 0.18094727515834508,
+      "learning_rate": 1.4683331167703218e-05,
+      "loss": 0.5502,
+      "step": 190
+    },
+    {
+      "epoch": 0.7276119402985075,
+      "grad_norm": 0.17080964382720645,
+      "learning_rate": 1.356492314681356e-05,
+      "loss": 0.553,
+      "step": 195
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 0.16164117491727123,
+      "learning_rate": 1.2499792556533716e-05,
+      "loss": 0.5474,
+      "step": 200
+    },
+    {
+      "epoch": 0.7649253731343284,
+      "grad_norm": 0.17787031389952734,
+      "learning_rate": 1.1492011668707753e-05,
+      "loss": 0.5449,
+      "step": 205
+    },
+    {
+      "epoch": 0.7835820895522388,
+      "grad_norm": 0.17503610601554878,
+      "learning_rate": 1.0545433492320603e-05,
+      "loss": 0.55,
+      "step": 210
+    },
+    {
+      "epoch": 0.8022388059701493,
+      "grad_norm": 0.16061415785978014,
+      "learning_rate": 9.663677042440537e-06,
+      "loss": 0.5443,
+      "step": 215
+    },
+    {
+      "epoch": 0.8208955223880597,
+      "grad_norm": 0.16426892962313516,
+      "learning_rate": 8.850113503781367e-06,
+      "loss": 0.5443,
+      "step": 220
+    },
+    {
+      "epoch": 0.8395522388059702,
+      "grad_norm": 0.14334733267655186,
+      "learning_rate": 8.107853341784671e-06,
+      "loss": 0.5506,
+      "step": 225
+    },
+    {
+      "epoch": 0.8582089552238806,
+      "grad_norm": 0.15691714417345864,
+      "learning_rate": 7.439734410499752e-06,
+      "loss": 0.547,
+      "step": 230
+    },
+    {
+      "epoch": 0.8768656716417911,
+      "grad_norm": 0.18311717197065572,
+      "learning_rate": 6.848311102728011e-06,
+      "loss": 0.5472,
+      "step": 235
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 0.18317039918731354,
+      "learning_rate": 6.335844583913515e-06,
+      "loss": 0.5433,
+      "step": 240
+    },
+    {
+      "epoch": 0.914179104477612,
+      "grad_norm": 0.14619058285902858,
+      "learning_rate": 5.904294147118193e-06,
+      "loss": 0.547,
+      "step": 245
+    },
+    {
+      "epoch": 0.9328358208955224,
+      "grad_norm": 0.14674904042247255,
+      "learning_rate": 5.555309722133842e-06,
+      "loss": 0.5435,
+      "step": 250
+    },
+    {
+      "epoch": 0.9514925373134329,
+      "grad_norm": 0.14162340991654054,
+      "learning_rate": 5.290225567370509e-06,
+      "loss": 0.5396,
+      "step": 255
+    },
+    {
+      "epoch": 0.9701492537313433,
+      "grad_norm": 0.14994290104981006,
+      "learning_rate": 5.110055168638854e-06,
+      "loss": 0.5433,
+      "step": 260
+    },
+    {
+      "epoch": 0.9888059701492538,
+      "grad_norm": 0.1429086930896721,
+      "learning_rate": 5.0154873643297575e-06,
+      "loss": 0.547,
+      "step": 265
+    },
+    {
+      "epoch": 1.0,
+      "step": 268,
+      "total_flos": 488165445992448.0,
+      "train_loss": 0.5792719663968727,
+      "train_runtime": 25010.1039,
+      "train_samples_per_second": 1.371,
+      "train_steps_per_second": 0.011
     }
   ],
   "logging_steps": 5,
-  "max_steps": 168,
+  "max_steps": 268,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
-  "save_steps": 500,
+  "save_steps": 100,
   "stateful_callbacks": {
     "TrainerControl": {
       "args": {
         "should_epoch_stop": false,
         "should_evaluate": false,
         "should_log": false,
-        "should_save": false,
-        "should_training_stop": false
+        "should_save": true,
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 76560728064000.0,
-  "train_batch_size": 2,
+  "total_flos": 488165445992448.0,
+  "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null
 }