Model save

README.md

@@ -1,11 +1,9 @@
 ---
 base_model: openai-community/gpt2
-datasets: HuggingFaceFW/fineweb-edu
 library_name: transformers
 model_name: gpt2-RMT-2
 tags:
 - generated_from_trainer
-- open-r1
 - trl
 - sft
 licence: license
@@ -13,7 +11,7 @@ licence: license
 
 # Model Card for gpt2-RMT-2
 
-This model is a fine-tuned version of [openai-community/gpt2](https://huggingface.co/openai-community/gpt2) on the [HuggingFaceFW/fineweb-edu](https://huggingface.co/datasets/HuggingFaceFW/fineweb-edu) dataset.
+This model is a fine-tuned version of [openai-community/gpt2](https://huggingface.co/openai-community/gpt2).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start
@@ -29,7 +27,7 @@ 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/s18574s18574-/huggingface/runs/mh05rq2b) 
+[<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/s18574s18574-/huggingface/runs/rtv4pjln) 
 
 
 This model was trained with SFT.

all_results.json

@@ -4,9 +4,9 @@
     "eval_samples_per_second": 194.329,
     "eval_steps_per_second": 12.146,
     "total_flos": 5419400756723712.0,
-    "train_loss": 3.6152698863636363,
-    "train_runtime": 337.2618,
+    "train_loss": 3.61015625,
+    "train_runtime": 338.5176,
     "train_samples": 19883,
-    "train_samples_per_second": 31.344,
-    "train_steps_per_second": 0.978
+    "train_samples_per_second": 31.227,
+    "train_steps_per_second": 0.975
 }

config.json

@@ -33,6 +33,6 @@
   },
   "torch_dtype": "bfloat16",
   "transformers_version": "4.50.0.dev0",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 50257
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b16e3936aa2c91fc567c45311766ae7a01909050a6b9a4ce29225967ebbf4bc
+oid sha256:1422a31ad724b5784e66d973fec1851ddb469e716474476e44603c91f8e30ca7
 size 248912768

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 5419400756723712.0,
-    "train_loss": 3.6152698863636363,
-    "train_runtime": 337.2618,
+    "train_loss": 3.61015625,
+    "train_runtime": 338.5176,
     "train_samples": 19883,
-    "train_samples_per_second": 31.344,
-    "train_steps_per_second": 0.978
+    "train_samples_per_second": 31.227,
+    "train_steps_per_second": 0.975
 }

trainer_state.json

@@ -10,297 +10,297 @@
   "log_history": [
     {
       "epoch": 0.030257186081694403,
-      "grad_norm": 51.5472786539882,
+      "grad_norm": 53.20387628855397,
       "learning_rate": 6.060606060606061e-06,
-      "loss": 7.3992,
-      "mean_token_accuracy": 0.13563802018761634,
+      "loss": 7.3984,
+      "mean_token_accuracy": 0.1354332946240902,
       "step": 10
     },
     {
       "epoch": 0.060514372163388806,
-      "grad_norm": 27.69300301659005,
+      "grad_norm": 27.17948871011392,
       "learning_rate": 1.2121212121212122e-05,
-      "loss": 6.8242,
-      "mean_token_accuracy": 0.14449115768074988,
+      "loss": 6.7727,
+      "mean_token_accuracy": 0.14481190741062164,
       "step": 20
     },
     {
       "epoch": 0.0907715582450832,
-      "grad_norm": 9.743705978929961,
+      "grad_norm": 9.47619461823253,
       "learning_rate": 1.8181818181818182e-05,
-      "loss": 5.6266,
-      "mean_token_accuracy": 0.1745789811015129,
+      "loss": 5.5984,
+      "mean_token_accuracy": 0.17588953599333762,
       "step": 30
     },
     {
       "epoch": 0.12102874432677761,
-      "grad_norm": 4.50812560927562,
+      "grad_norm": 4.801175322490429,
       "learning_rate": 1.9972599751485225e-05,
-      "loss": 4.6586,
-      "mean_token_accuracy": 0.24351915791630746,
+      "loss": 4.6516,
+      "mean_token_accuracy": 0.24421471059322358,
       "step": 40
     },
     {
       "epoch": 0.15128593040847202,
-      "grad_norm": 3.1996689209062676,
+      "grad_norm": 3.381736221106582,
       "learning_rate": 1.9838755799290993e-05,
-      "loss": 4.0187,
-      "mean_token_accuracy": 0.3023824915289879,
+      "loss": 4.009,
+      "mean_token_accuracy": 0.3037181183695793,
       "step": 50
     },
     {
       "epoch": 0.1815431164901664,
-      "grad_norm": 1.8824084218377581,
+      "grad_norm": 1.8196128883676905,
       "learning_rate": 1.9594929736144978e-05,
-      "loss": 3.6316,
-      "mean_token_accuracy": 0.3426647365093231,
+      "loss": 3.5969,
+      "mean_token_accuracy": 0.3466367706656456,
       "step": 60
     },
     {
       "epoch": 0.2118003025718608,
-      "grad_norm": 1.1187763043308236,
+      "grad_norm": 1.0989833139359242,
       "learning_rate": 1.9243847161266924e-05,
-      "loss": 3.3883,
-      "mean_token_accuracy": 0.3721240028738976,
+      "loss": 3.3781,
+      "mean_token_accuracy": 0.37338929772377016,
       "step": 70
     },
     {
       "epoch": 0.24205748865355523,
-      "grad_norm": 1.0093619517882768,
+      "grad_norm": 0.9673089083213241,
       "learning_rate": 1.8789432636206197e-05,
-      "loss": 3.3281,
-      "mean_token_accuracy": 0.3791012689471245,
+      "loss": 3.325,
+      "mean_token_accuracy": 0.37947496473789216,
       "step": 80
     },
     {
       "epoch": 0.2723146747352496,
-      "grad_norm": 0.8420416221184239,
+      "grad_norm": 0.8526479435566983,
       "learning_rate": 1.8236765814298328e-05,
-      "loss": 3.2738,
-      "mean_token_accuracy": 0.3867510572075844,
+      "loss": 3.2707,
+      "mean_token_accuracy": 0.3870904952287674,
       "step": 90
     },
     {
       "epoch": 0.30257186081694404,
-      "grad_norm": 0.8952752980575562,
+      "grad_norm": 0.8771420630730227,
       "learning_rate": 1.7592024657977432e-05,
-      "loss": 3.2816,
-      "mean_token_accuracy": 0.3850349023938179,
+      "loss": 3.2793,
+      "mean_token_accuracy": 0.3854751497507095,
       "step": 100
     },
     {
       "epoch": 0.30257186081694404,
-      "eval_runtime": 0.2249,
-      "eval_samples_per_second": 213.425,
-      "eval_steps_per_second": 13.339,
+      "eval_runtime": 0.2258,
+      "eval_samples_per_second": 212.573,
+      "eval_steps_per_second": 13.286,
       "step": 100
     },
     {
       "epoch": 0.3328290468986384,
-      "grad_norm": 0.8146530137770093,
+      "grad_norm": 0.7981055028496323,
       "learning_rate": 1.686241637868734e-05,
-      "loss": 3.2516,
-      "mean_token_accuracy": 0.3902030408382416,
+      "loss": 3.2504,
+      "mean_token_accuracy": 0.3904677391052246,
       "step": 110
     },
     {
       "epoch": 0.3630862329803328,
-      "grad_norm": 0.7245362247896451,
+      "grad_norm": 0.7170465908499676,
       "learning_rate": 1.6056096871376667e-05,
-      "loss": 3.2777,
-      "mean_token_accuracy": 0.3867401584982872,
+      "loss": 3.277,
+      "mean_token_accuracy": 0.3867432788014412,
       "step": 120
     },
     {
       "epoch": 0.39334341906202724,
-      "grad_norm": 0.7046871234836425,
+      "grad_norm": 0.7497925109759247,
       "learning_rate": 1.5182079544175957e-05,
-      "loss": 3.25,
-      "mean_token_accuracy": 0.39042258113622663,
+      "loss": 3.2488,
+      "mean_token_accuracy": 0.39058607518672944,
       "step": 130
     },
     {
       "epoch": 0.4236006051437216,
-      "grad_norm": 0.7889558903633703,
+      "grad_norm": 0.7800876699148922,
       "learning_rate": 1.4250134562400301e-05,
-      "loss": 3.2254,
-      "mean_token_accuracy": 0.3929138645529747,
+      "loss": 3.2242,
+      "mean_token_accuracy": 0.392949678003788,
       "step": 140
     },
     {
       "epoch": 0.45385779122541603,
-      "grad_norm": 0.7526789075759445,
+      "grad_norm": 0.7636894413901015,
       "learning_rate": 1.3270679633174219e-05,
-      "loss": 3.2168,
-      "mean_token_accuracy": 0.3945690140128136,
+      "loss": 3.2141,
+      "mean_token_accuracy": 0.3948305919766426,
       "step": 150
     },
     {
       "epoch": 0.48411497730711045,
-      "grad_norm": 0.7037431923979744,
+      "grad_norm": 0.7028258888259162,
       "learning_rate": 1.2254663551538047e-05,
-      "loss": 3.1898,
-      "mean_token_accuracy": 0.39688222110271454,
+      "loss": 3.1902,
+      "mean_token_accuracy": 0.39704713225364685,
       "step": 160
     },
     {
       "epoch": 0.5143721633888049,
-      "grad_norm": 0.7269697505833069,
+      "grad_norm": 0.7254535951054281,
       "learning_rate": 1.121344380981082e-05,
       "loss": 3.2254,
-      "mean_token_accuracy": 0.39344793409109113,
+      "mean_token_accuracy": 0.3936114251613617,
       "step": 170
     },
     {
       "epoch": 0.5446293494704992,
-      "grad_norm": 0.71564273988106,
+      "grad_norm": 0.7313179656460369,
       "learning_rate": 1.015865963834808e-05,
-      "loss": 3.198,
-      "mean_token_accuracy": 0.3971615001559258,
+      "loss": 3.1965,
+      "mean_token_accuracy": 0.39705405980348585,
       "step": 180
     },
     {
       "epoch": 0.5748865355521936,
-      "grad_norm": 0.7272400173576803,
+      "grad_norm": 0.6925411391649295,
       "learning_rate": 9.102101896903084e-06,
-      "loss": 3.2098,
-      "mean_token_accuracy": 0.3965013727545738,
+      "loss": 3.2078,
+      "mean_token_accuracy": 0.3967175245285034,
       "step": 190
     },
     {
       "epoch": 0.6051437216338881,
-      "grad_norm": 0.7204839560593628,
+      "grad_norm": 0.7015683467230908,
       "learning_rate": 8.055581271005292e-06,
       "loss": 3.209,
-      "mean_token_accuracy": 0.394715978205204,
+      "mean_token_accuracy": 0.39455645680427553,
       "step": 200
     },
     {
       "epoch": 0.6051437216338881,
-      "eval_runtime": 0.2231,
-      "eval_samples_per_second": 215.144,
-      "eval_steps_per_second": 13.447,
+      "eval_runtime": 0.2238,
+      "eval_samples_per_second": 214.446,
+      "eval_steps_per_second": 13.403,
       "step": 200
     },
     {
       "epoch": 0.6354009077155824,
-      "grad_norm": 0.6620216126492257,
+      "grad_norm": 0.726599375455085,
       "learning_rate": 7.0307962467172555e-06,
-      "loss": 3.1988,
-      "mean_token_accuracy": 0.39781390130519867,
+      "loss": 3.1992,
+      "mean_token_accuracy": 0.3978341415524483,
       "step": 210
     },
     {
       "epoch": 0.6656580937972768,
-      "grad_norm": 0.7813459496424706,
+      "grad_norm": 0.6917092811454406,
       "learning_rate": 6.039202339608432e-06,
       "loss": 3.202,
-      "mean_token_accuracy": 0.3953568756580353,
+      "mean_token_accuracy": 0.3954534113407135,
       "step": 220
     },
     {
       "epoch": 0.6959152798789713,
-      "grad_norm": 0.6556046091666947,
+      "grad_norm": 0.6705832721497097,
       "learning_rate": 5.091884039764321e-06,
       "loss": 3.1969,
-      "mean_token_accuracy": 0.3961416259407997,
+      "mean_token_accuracy": 0.3959610104560852,
       "step": 230
     },
     {
       "epoch": 0.7261724659606656,
-      "grad_norm": 0.7243524981690901,
+      "grad_norm": 0.7043055334462106,
       "learning_rate": 4.19943090428802e-06,
-      "loss": 3.2102,
-      "mean_token_accuracy": 0.3935989648103714,
+      "loss": 3.2098,
+      "mean_token_accuracy": 0.39361765086650846,
       "step": 240
     },
     {
       "epoch": 0.75642965204236,
-      "grad_norm": 0.7423682208710032,
+      "grad_norm": 0.7337550100582009,
       "learning_rate": 3.37181918238904e-06,
-      "loss": 3.2078,
-      "mean_token_accuracy": 0.39539735913276675,
+      "loss": 3.207,
+      "mean_token_accuracy": 0.3955203667283058,
       "step": 250
     },
     {
       "epoch": 0.7866868381240545,
-      "grad_norm": 0.684794452564992,
+      "grad_norm": 0.6893800072740404,
       "learning_rate": 2.618300296308135e-06,
-      "loss": 3.2219,
-      "mean_token_accuracy": 0.3939197212457657,
+      "loss": 3.2207,
+      "mean_token_accuracy": 0.3941065609455109,
       "step": 260
     },
     {
       "epoch": 0.8169440242057489,
-      "grad_norm": 0.6857522826014064,
+      "grad_norm": 0.674883610377537,
       "learning_rate": 1.947297424689414e-06,
-      "loss": 3.1992,
-      "mean_token_accuracy": 0.39573368430137634,
+      "loss": 3.1988,
+      "mean_token_accuracy": 0.3957181140780449,
       "step": 270
     },
     {
       "epoch": 0.8472012102874432,
-      "grad_norm": 0.6966190236263456,
+      "grad_norm": 0.6757567363771522,
       "learning_rate": 1.3663113444380905e-06,
-      "loss": 3.2031,
-      "mean_token_accuracy": 0.3933654069900513,
+      "loss": 3.2012,
+      "mean_token_accuracy": 0.39344170689582825,
       "step": 280
     },
     {
       "epoch": 0.8774583963691377,
-      "grad_norm": 0.8226214667450941,
+      "grad_norm": 0.7166978012385395,
       "learning_rate": 8.818365836066101e-07,
       "loss": 3.1621,
-      "mean_token_accuracy": 0.4012774467468262,
+      "mean_token_accuracy": 0.4014370396733284,
       "step": 290
     },
     {
       "epoch": 0.9077155824508321,
-      "grad_norm": 0.6811554548251312,
+      "grad_norm": 0.671285323335886,
       "learning_rate": 4.992888225905467e-07,
-      "loss": 3.2254,
-      "mean_token_accuracy": 0.39436970353126527,
+      "loss": 3.2242,
+      "mean_token_accuracy": 0.3944755882024765,
       "step": 300
     },
     {
       "epoch": 0.9077155824508321,
-      "eval_runtime": 0.223,
-      "eval_samples_per_second": 215.211,
-      "eval_steps_per_second": 13.451,
+      "eval_runtime": 0.2237,
+      "eval_samples_per_second": 214.554,
+      "eval_steps_per_second": 13.41,
       "step": 300
     },
     {
       "epoch": 0.9379727685325264,
-      "grad_norm": 0.6793742171946109,
+      "grad_norm": 0.6982272552320608,
       "learning_rate": 2.2294435517691504e-07,
-      "loss": 3.1926,
-      "mean_token_accuracy": 0.397060464322567,
+      "loss": 3.1918,
+      "mean_token_accuracy": 0.39713545739650724,
       "step": 310
     },
     {
       "epoch": 0.9682299546142209,
-      "grad_norm": 0.6744479467043031,
+      "grad_norm": 0.6635317905732051,
       "learning_rate": 5.5892286176932875e-08,
-      "loss": 3.1961,
-      "mean_token_accuracy": 0.3960541605949402,
+      "loss": 3.1957,
+      "mean_token_accuracy": 0.3961677893996239,
       "step": 320
     },
     {
       "epoch": 0.9984871406959153,
-      "grad_norm": 0.6621754518709608,
+      "grad_norm": 0.6569802422298082,
       "learning_rate": 0.0,
-      "loss": 3.2035,
-      "mean_token_accuracy": 0.3947854354977608,
+      "loss": 3.2023,
+      "mean_token_accuracy": 0.3949317991733551,
       "step": 330
     },
     {
       "epoch": 0.9984871406959153,
       "step": 330,
       "total_flos": 5419400756723712.0,
-      "train_loss": 3.6152698863636363,
-      "train_runtime": 337.2618,
-      "train_samples_per_second": 31.344,
-      "train_steps_per_second": 0.978
+      "train_loss": 3.61015625,
+      "train_runtime": 338.5176,
+      "train_samples_per_second": 31.227,
+      "train_steps_per_second": 0.975
     }
   ],
   "logging_steps": 10,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0dba648ef7d2addc48fb53ca1ab6b041442778aef087f173aef4c1a8ab1b68d9
+oid sha256:3c63eefca7f89a43cb29b332e3e763cfdc0b88b24fde057b72ff9133306fc667
 size 7352