Model save

README.md

@@ -4,7 +4,6 @@ library_name: transformers
 model_name: llama_tuned
 tags:
 - generated_from_trainer
-- alignment-handbook
 - trl
 - sft
 licence: license
@@ -28,7 +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/vedaantj/ft-llm/runs/bg1fm47k)
+[<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/vedaantj/ft-llm/runs/00gc56v6)
 
 This model was trained with SFT.
 

all_results.json

@@ -1,14 +1,14 @@
 {
-    "epoch": 20.0,
+    "epoch": 40.0,
     "eval_loss": 6.792967319488525,
     "eval_runtime": 0.1698,
     "eval_samples": 10,
     "eval_samples_per_second": 58.885,
     "eval_steps_per_second": 11.777,
-    "total_flos": 2482684999761920.0,
-    "train_loss": 8.701183927059173,
-    "train_runtime": 15.7189,
+    "total_flos": 4882092898910208.0,
+    "train_loss": 0.5672883474326227,
+    "train_runtime": 31.3668,
     "train_samples": 10,
-    "train_samples_per_second": 12.724,
-    "train_steps_per_second": 1.272
+    "train_samples_per_second": 12.752,
+    "train_steps_per_second": 1.275
 }

config.json

@@ -36,6 +36,6 @@
   "tie_word_embeddings": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.46.3",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 128256
 }

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:047f6c5d065c8397487a2cdf995de35a63777d1107881ee0882d634f051b7ef0
+oid sha256:633bf1ca74fc04074527717859d80919eccf368ed25c2972229a37d4ee376b11
 size 4965799096

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b926f18baa51bd6f8c64587045629923fd6317fce0ae32eca72d2b5a218178b
+oid sha256:67bb5340c0a30c7b09d3e885637cc10a84a99e891ddc1b37a3b2ca062d7e5c23
 size 2247734992

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 20.0,
-    "total_flos": 2482684999761920.0,
-    "train_loss": 8.701183927059173,
-    "train_runtime": 15.7189,
+    "epoch": 40.0,
+    "total_flos": 4882092898910208.0,
+    "train_loss": 0.5672883474326227,
+    "train_runtime": 31.3668,
     "train_samples": 10,
-    "train_samples_per_second": 12.724,
-    "train_steps_per_second": 1.272
+    "train_samples_per_second": 12.752,
+    "train_steps_per_second": 1.275
 }

trainer_state.json

@@ -1,326 +1,626 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 20.0,
+  "epoch": 40.0,
   "eval_steps": 500,
-  "global_step": 20,
+  "global_step": 40,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002,
+      "learning_rate": 2.5e-05,
       "loss": 1.1277,
       "step": 1
     },
     {
       "epoch": 1.0,
       "eval_loss": 1.131566047668457,
-      "eval_runtime": 0.1562,
-      "eval_samples_per_second": 64.008,
-      "eval_steps_per_second": 12.802,
+      "eval_runtime": 0.1598,
+      "eval_samples_per_second": 62.597,
+      "eval_steps_per_second": 12.519,
       "step": 1
     },
     {
       "epoch": 2.0,
-      "grad_norm": 9.75125789642334,
-      "learning_rate": 0.0002,
+      "grad_norm": 9.749205589294434,
+      "learning_rate": 5e-05,
       "loss": 1.1065,
       "step": 2
     },
     {
       "epoch": 2.0,
-      "eval_loss": 3.3900864124298096,
-      "eval_runtime": 0.1564,
-      "eval_samples_per_second": 63.93,
-      "eval_steps_per_second": 12.786,
+      "eval_loss": 0.7347294092178345,
+      "eval_runtime": 0.1573,
+      "eval_samples_per_second": 63.584,
+      "eval_steps_per_second": 12.717,
       "step": 2
     },
     {
       "epoch": 3.0,
-      "grad_norm": 9.75125789642334,
-      "learning_rate": 0.0002,
-      "loss": 3.4811,
+      "grad_norm": 9.749205589294434,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.7329,
       "step": 3
     },
     {
       "epoch": 3.0,
-      "eval_loss": 3.3900864124298096,
-      "eval_runtime": 0.1554,
-      "eval_samples_per_second": 64.349,
-      "eval_steps_per_second": 12.87,
+      "eval_loss": 0.7347294092178345,
+      "eval_runtime": 0.1562,
+      "eval_samples_per_second": 64.01,
+      "eval_steps_per_second": 12.802,
       "step": 3
     },
     {
       "epoch": 4.0,
-      "grad_norm": 36.552181243896484,
-      "learning_rate": 0.0002,
-      "loss": 3.4725,
+      "grad_norm": 6.555100440979004,
+      "learning_rate": 0.0001,
+      "loss": 0.7406,
       "step": 4
     },
     {
       "epoch": 4.0,
-      "eval_loss": 11.004281997680664,
-      "eval_runtime": 0.1554,
-      "eval_samples_per_second": 64.37,
-      "eval_steps_per_second": 12.874,
+      "eval_loss": 1.1057980060577393,
+      "eval_runtime": 0.1567,
+      "eval_samples_per_second": 63.83,
+      "eval_steps_per_second": 12.766,
       "step": 4
     },
     {
       "epoch": 5.0,
-      "grad_norm": 36.552181243896484,
-      "learning_rate": 0.0002,
-      "loss": 11.0104,
+      "grad_norm": 6.555100440979004,
+      "learning_rate": 0.000125,
+      "loss": 1.0826,
       "step": 5
     },
     {
       "epoch": 5.0,
-      "eval_loss": 11.004281997680664,
-      "eval_runtime": 0.1554,
-      "eval_samples_per_second": 64.354,
-      "eval_steps_per_second": 12.871,
+      "eval_loss": 1.1057980060577393,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.313,
+      "eval_steps_per_second": 12.863,
       "step": 5
     },
     {
       "epoch": 6.0,
-      "grad_norm": 200.41488647460938,
-      "learning_rate": 0.0002,
-      "loss": 10.9014,
+      "grad_norm": 18.290578842163086,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.0478,
       "step": 6
     },
     {
       "epoch": 6.0,
-      "eval_loss": 13.01452922821045,
-      "eval_runtime": 0.1558,
-      "eval_samples_per_second": 64.169,
-      "eval_steps_per_second": 12.834,
+      "eval_loss": 1.3313286304473877,
+      "eval_runtime": 0.1563,
+      "eval_samples_per_second": 63.987,
+      "eval_steps_per_second": 12.797,
       "step": 6
     },
     {
       "epoch": 7.0,
-      "grad_norm": 200.41488647460938,
-      "learning_rate": 0.0002,
-      "loss": 13.007,
+      "grad_norm": 18.290578842163086,
+      "learning_rate": 0.000175,
+      "loss": 1.3493,
       "step": 7
     },
     {
       "epoch": 7.0,
-      "eval_loss": 13.01452922821045,
-      "eval_runtime": 0.1562,
-      "eval_samples_per_second": 64.025,
-      "eval_steps_per_second": 12.805,
+      "eval_loss": 1.3313286304473877,
+      "eval_runtime": 0.1554,
+      "eval_samples_per_second": 64.364,
+      "eval_steps_per_second": 12.873,
       "step": 7
     },
     {
       "epoch": 8.0,
-      "grad_norm": 56.3553581237793,
+      "grad_norm": 51.25260925292969,
       "learning_rate": 0.0002,
-      "loss": 13.0271,
+      "loss": 1.4307,
       "step": 8
     },
     {
       "epoch": 8.0,
-      "eval_loss": 10.359188079833984,
-      "eval_runtime": 0.155,
-      "eval_samples_per_second": 64.517,
-      "eval_steps_per_second": 12.903,
+      "eval_loss": 0.7719938158988953,
+      "eval_runtime": 0.1571,
+      "eval_samples_per_second": 63.674,
+      "eval_steps_per_second": 12.735,
       "step": 8
     },
     {
       "epoch": 9.0,
-      "grad_norm": 56.3553581237793,
-      "learning_rate": 0.0002,
-      "loss": 10.5514,
+      "grad_norm": 51.25260925292969,
+      "learning_rate": 0.0001995184726672197,
+      "loss": 0.7569,
       "step": 9
     },
     {
       "epoch": 9.0,
-      "eval_loss": 10.359188079833984,
-      "eval_runtime": 0.17,
-      "eval_samples_per_second": 58.815,
-      "eval_steps_per_second": 11.763,
+      "eval_loss": 0.7719938158988953,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.326,
+      "eval_steps_per_second": 12.865,
       "step": 9
     },
     {
       "epoch": 10.0,
-      "grad_norm": 37.37253189086914,
-      "learning_rate": 0.0002,
-      "loss": 10.6183,
+      "grad_norm": 7.65546178817749,
+      "learning_rate": 0.00019807852804032305,
+      "loss": 0.785,
       "step": 10
     },
     {
       "epoch": 10.0,
-      "eval_loss": 16.391109466552734,
-      "eval_runtime": 0.1578,
-      "eval_samples_per_second": 63.388,
-      "eval_steps_per_second": 12.678,
+      "eval_loss": 3.317613124847412,
+      "eval_runtime": 0.1563,
+      "eval_samples_per_second": 63.973,
+      "eval_steps_per_second": 12.795,
       "step": 10
     },
     {
       "epoch": 11.0,
-      "grad_norm": 37.37253189086914,
-      "learning_rate": 0.0002,
-      "loss": 16.3831,
+      "grad_norm": 7.65546178817749,
+      "learning_rate": 0.0001956940335732209,
+      "loss": 3.3605,
       "step": 11
     },
     {
       "epoch": 11.0,
-      "eval_loss": 16.391109466552734,
-      "eval_runtime": 0.1603,
-      "eval_samples_per_second": 62.396,
-      "eval_steps_per_second": 12.479,
+      "eval_loss": 3.317613124847412,
+      "eval_runtime": 0.1553,
+      "eval_samples_per_second": 64.372,
+      "eval_steps_per_second": 12.874,
       "step": 11
     },
     {
       "epoch": 12.0,
-      "grad_norm": 25.48611068725586,
-      "learning_rate": 0.0002,
-      "loss": 16.3559,
+      "grad_norm": 132.75279235839844,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 3.4664,
       "step": 12
     },
     {
       "epoch": 12.0,
-      "eval_loss": 8.423013687133789,
-      "eval_runtime": 0.1556,
-      "eval_samples_per_second": 64.286,
-      "eval_steps_per_second": 12.857,
+      "eval_loss": 1.0249794721603394,
+      "eval_runtime": 0.1586,
+      "eval_samples_per_second": 63.039,
+      "eval_steps_per_second": 12.608,
       "step": 12
     },
     {
       "epoch": 13.0,
-      "grad_norm": 25.48611068725586,
-      "learning_rate": 0.0002,
-      "loss": 8.5027,
+      "grad_norm": 132.75279235839844,
+      "learning_rate": 0.0001881921264348355,
+      "loss": 1.0922,
       "step": 13
     },
     {
       "epoch": 13.0,
-      "eval_loss": 8.423013687133789,
-      "eval_runtime": 0.1577,
-      "eval_samples_per_second": 63.415,
-      "eval_steps_per_second": 12.683,
+      "eval_loss": 1.0249794721603394,
+      "eval_runtime": 0.1558,
+      "eval_samples_per_second": 64.204,
+      "eval_steps_per_second": 12.841,
       "step": 13
     },
     {
       "epoch": 14.0,
-      "grad_norm": 22.36524772644043,
-      "learning_rate": 0.0002,
-      "loss": 8.4487,
+      "grad_norm": 10.862001419067383,
+      "learning_rate": 0.00018314696123025454,
+      "loss": 1.0792,
       "step": 14
     },
     {
       "epoch": 14.0,
-      "eval_loss": 8.201436042785645,
-      "eval_runtime": 0.1568,
-      "eval_samples_per_second": 63.761,
-      "eval_steps_per_second": 12.752,
+      "eval_loss": 0.7917268872261047,
+      "eval_runtime": 0.1564,
+      "eval_samples_per_second": 63.95,
+      "eval_steps_per_second": 12.79,
       "step": 14
     },
     {
       "epoch": 15.0,
-      "grad_norm": 22.36524772644043,
-      "learning_rate": 0.0002,
-      "loss": 8.2898,
+      "grad_norm": 10.862001419067383,
+      "learning_rate": 0.0001773010453362737,
+      "loss": 0.8238,
       "step": 15
     },
     {
       "epoch": 15.0,
-      "eval_loss": 8.201436042785645,
-      "eval_runtime": 0.159,
-      "eval_samples_per_second": 62.881,
-      "eval_steps_per_second": 12.576,
+      "eval_loss": 0.7917268872261047,
+      "eval_runtime": 0.1557,
+      "eval_samples_per_second": 64.219,
+      "eval_steps_per_second": 12.844,
       "step": 15
     },
     {
       "epoch": 16.0,
-      "grad_norm": 24.524402618408203,
-      "learning_rate": 0.0002,
-      "loss": 8.2531,
+      "grad_norm": 20.844392776489258,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.8054,
       "step": 16
     },
     {
       "epoch": 16.0,
-      "eval_loss": 7.470841407775879,
-      "eval_runtime": 0.1568,
-      "eval_samples_per_second": 63.756,
-      "eval_steps_per_second": 12.751,
+      "eval_loss": 0.3446378707885742,
+      "eval_runtime": 0.1553,
+      "eval_samples_per_second": 64.409,
+      "eval_steps_per_second": 12.882,
       "step": 16
     },
     {
       "epoch": 17.0,
-      "grad_norm": 24.524402618408203,
-      "learning_rate": 0.0002,
-      "loss": 7.5485,
+      "grad_norm": 20.844392776489258,
+      "learning_rate": 0.00016343932841636456,
+      "loss": 0.3504,
       "step": 17
     },
     {
       "epoch": 17.0,
-      "eval_loss": 7.470841407775879,
-      "eval_runtime": 0.1593,
-      "eval_samples_per_second": 62.762,
-      "eval_steps_per_second": 12.552,
+      "eval_loss": 0.3446378707885742,
+      "eval_runtime": 0.157,
+      "eval_samples_per_second": 63.679,
+      "eval_steps_per_second": 12.736,
       "step": 17
     },
     {
       "epoch": 18.0,
-      "grad_norm": 14.851771354675293,
-      "learning_rate": 0.0002,
-      "loss": 7.4808,
+      "grad_norm": 3.432838201522827,
+      "learning_rate": 0.00015555702330196023,
+      "loss": 0.3424,
       "step": 18
     },
     {
       "epoch": 18.0,
-      "eval_loss": 7.270651817321777,
-      "eval_runtime": 0.1579,
-      "eval_samples_per_second": 63.312,
-      "eval_steps_per_second": 12.662,
+      "eval_loss": 0.27769702672958374,
+      "eval_runtime": 0.1557,
+      "eval_samples_per_second": 64.21,
+      "eval_steps_per_second": 12.842,
       "step": 18
     },
     {
       "epoch": 19.0,
-      "grad_norm": 14.851771354675293,
-      "learning_rate": 0.0002,
-      "loss": 7.2185,
+      "grad_norm": 3.432838201522827,
+      "learning_rate": 0.0001471396736825998,
+      "loss": 0.2479,
       "step": 19
     },
     {
       "epoch": 19.0,
-      "eval_loss": 7.270651817321777,
-      "eval_runtime": 0.1576,
-      "eval_samples_per_second": 63.465,
-      "eval_steps_per_second": 12.693,
+      "eval_loss": 0.27769702672958374,
+      "eval_runtime": 0.1567,
+      "eval_samples_per_second": 63.823,
+      "eval_steps_per_second": 12.765,
       "step": 19
     },
     {
       "epoch": 20.0,
-      "grad_norm": 11.238504409790039,
-      "learning_rate": 0.0002,
-      "loss": 7.239,
+      "grad_norm": 2.582324981689453,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.244,
       "step": 20
     },
     {
       "epoch": 20.0,
-      "eval_loss": 6.792967319488525,
-      "eval_runtime": 0.1595,
-      "eval_samples_per_second": 62.7,
-      "eval_steps_per_second": 12.54,
+      "eval_loss": 0.1163192018866539,
+      "eval_runtime": 0.1566,
+      "eval_samples_per_second": 63.838,
+      "eval_steps_per_second": 12.768,
       "step": 20
     },
     {
-      "epoch": 20.0,
-      "step": 20,
-      "total_flos": 2482684999761920.0,
-      "train_loss": 8.701183927059173,
-      "train_runtime": 15.7189,
-      "train_samples_per_second": 12.724,
-      "train_steps_per_second": 1.272
+      "epoch": 21.0,
+      "grad_norm": 2.582324981689453,
+      "learning_rate": 0.00012902846772544624,
+      "loss": 0.1097,
+      "step": 21
+    },
+    {
+      "epoch": 21.0,
+      "eval_loss": 0.1163192018866539,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.297,
+      "eval_steps_per_second": 12.859,
+      "step": 21
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 1.2364883422851562,
+      "learning_rate": 0.00011950903220161285,
+      "loss": 0.1126,
+      "step": 22
+    },
+    {
+      "epoch": 22.0,
+      "eval_loss": 0.0716390460729599,
+      "eval_runtime": 0.1561,
+      "eval_samples_per_second": 64.064,
+      "eval_steps_per_second": 12.813,
+      "step": 22
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 1.2364883422851562,
+      "learning_rate": 0.0001098017140329561,
+      "loss": 0.0797,
+      "step": 23
+    },
+    {
+      "epoch": 23.0,
+      "eval_loss": 0.0716390460729599,
+      "eval_runtime": 0.1579,
+      "eval_samples_per_second": 63.343,
+      "eval_steps_per_second": 12.669,
+      "step": 23
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 1.081320881843567,
+      "learning_rate": 0.0001,
+      "loss": 0.0669,
+      "step": 24
+    },
+    {
+      "epoch": 24.0,
+      "eval_loss": 0.05489348620176315,
+      "eval_runtime": 0.1558,
+      "eval_samples_per_second": 64.175,
+      "eval_steps_per_second": 12.835,
+      "step": 24
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 1.081320881843567,
+      "learning_rate": 9.019828596704394e-05,
+      "loss": 0.0561,
+      "step": 25
+    },
+    {
+      "epoch": 25.0,
+      "eval_loss": 0.05489348620176315,
+      "eval_runtime": 0.1575,
+      "eval_samples_per_second": 63.477,
+      "eval_steps_per_second": 12.695,
+      "step": 25
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 1.572256326675415,
+      "learning_rate": 8.049096779838719e-05,
+      "loss": 0.0569,
+      "step": 26
+    },
+    {
+      "epoch": 26.0,
+      "eval_loss": 0.03670552000403404,
+      "eval_runtime": 0.1561,
+      "eval_samples_per_second": 64.056,
+      "eval_steps_per_second": 12.811,
+      "step": 26
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 1.572256326675415,
+      "learning_rate": 7.097153227455379e-05,
+      "loss": 0.0357,
+      "step": 27
+    },
+    {
+      "epoch": 27.0,
+      "eval_loss": 0.03670552000403404,
+      "eval_runtime": 0.1559,
+      "eval_samples_per_second": 64.161,
+      "eval_steps_per_second": 12.832,
+      "step": 27
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 0.9304390549659729,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.0331,
+      "step": 28
+    },
+    {
+      "epoch": 28.0,
+      "eval_loss": 0.05289148539304733,
+      "eval_runtime": 0.1561,
+      "eval_samples_per_second": 64.054,
+      "eval_steps_per_second": 12.811,
+      "step": 28
+    },
+    {
+      "epoch": 29.0,
+      "grad_norm": 0.9304390549659729,
+      "learning_rate": 5.286032631740023e-05,
+      "loss": 0.0454,
+      "step": 29
+    },
+    {
+      "epoch": 29.0,
+      "eval_loss": 0.05289148539304733,
+      "eval_runtime": 0.1561,
+      "eval_samples_per_second": 64.068,
+      "eval_steps_per_second": 12.814,
+      "step": 29
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 3.730225086212158,
+      "learning_rate": 4.444297669803981e-05,
+      "loss": 0.0562,
+      "step": 30
+    },
+    {
+      "epoch": 30.0,
+      "eval_loss": 0.02020808681845665,
+      "eval_runtime": 0.1578,
+      "eval_samples_per_second": 63.372,
+      "eval_steps_per_second": 12.674,
+      "step": 30
+    },
+    {
+      "epoch": 31.0,
+      "grad_norm": 3.730225086212158,
+      "learning_rate": 3.6560671583635467e-05,
+      "loss": 0.0184,
+      "step": 31
+    },
+    {
+      "epoch": 31.0,
+      "eval_loss": 0.02020808681845665,
+      "eval_runtime": 0.156,
+      "eval_samples_per_second": 64.11,
+      "eval_steps_per_second": 12.822,
+      "step": 31
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 0.8519256711006165,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.0186,
+      "step": 32
+    },
+    {
+      "epoch": 32.0,
+      "eval_loss": 0.006217154674232006,
+      "eval_runtime": 0.1564,
+      "eval_samples_per_second": 63.957,
+      "eval_steps_per_second": 12.791,
+      "step": 32
+    },
+    {
+      "epoch": 33.0,
+      "grad_norm": 0.8519256711006165,
+      "learning_rate": 2.26989546637263e-05,
+      "loss": 0.0056,
+      "step": 33
+    },
+    {
+      "epoch": 33.0,
+      "eval_loss": 0.006217154674232006,
+      "eval_runtime": 0.1582,
+      "eval_samples_per_second": 63.206,
+      "eval_steps_per_second": 12.641,
+      "step": 33
+    },
+    {
+      "epoch": 34.0,
+      "grad_norm": 0.17823612689971924,
+      "learning_rate": 1.6853038769745467e-05,
+      "loss": 0.0055,
+      "step": 34
+    },
+    {
+      "epoch": 34.0,
+      "eval_loss": 0.004052485339343548,
+      "eval_runtime": 0.1567,
+      "eval_samples_per_second": 63.821,
+      "eval_steps_per_second": 12.764,
+      "step": 34
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 0.17823612689971924,
+      "learning_rate": 1.1807873565164506e-05,
+      "loss": 0.0038,
+      "step": 35
+    },
+    {
+      "epoch": 35.0,
+      "eval_loss": 0.004052485339343548,
+      "eval_runtime": 0.1558,
+      "eval_samples_per_second": 64.205,
+      "eval_steps_per_second": 12.841,
+      "step": 35
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 0.14803585410118103,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 0.0037,
+      "step": 36
+    },
+    {
+      "epoch": 36.0,
+      "eval_loss": 0.0032847567927092314,
+      "eval_runtime": 0.1562,
+      "eval_samples_per_second": 64.036,
+      "eval_steps_per_second": 12.807,
+      "step": 36
+    },
+    {
+      "epoch": 37.0,
+      "grad_norm": 0.14803585410118103,
+      "learning_rate": 4.305966426779118e-06,
+      "loss": 0.0031,
+      "step": 37
+    },
+    {
+      "epoch": 37.0,
+      "eval_loss": 0.0032847567927092314,
+      "eval_runtime": 0.1553,
+      "eval_samples_per_second": 64.378,
+      "eval_steps_per_second": 12.876,
+      "step": 37
+    },
+    {
+      "epoch": 38.0,
+      "grad_norm": 0.08926349878311157,
+      "learning_rate": 1.921471959676957e-06,
+      "loss": 0.0028,
+      "step": 38
+    },
+    {
+      "epoch": 38.0,
+      "eval_loss": 0.0030998815782368183,
+      "eval_runtime": 0.1577,
+      "eval_samples_per_second": 63.417,
+      "eval_steps_per_second": 12.683,
+      "step": 38
+    },
+    {
+      "epoch": 39.0,
+      "grad_norm": 0.08926349878311157,
+      "learning_rate": 4.815273327803182e-07,
+      "loss": 0.0028,
+      "step": 39
+    },
+    {
+      "epoch": 39.0,
+      "eval_loss": 0.0030998815782368183,
+      "eval_runtime": 0.1568,
+      "eval_samples_per_second": 63.793,
+      "eval_steps_per_second": 12.759,
+      "step": 39
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 0.08725006133317947,
+      "learning_rate": 0.0,
+      "loss": 0.0028,
+      "step": 40
+    },
+    {
+      "epoch": 40.0,
+      "eval_loss": 0.0030517119448632,
+      "eval_runtime": 0.1564,
+      "eval_samples_per_second": 63.934,
+      "eval_steps_per_second": 12.787,
+      "step": 40
+    },
+    {
+      "epoch": 40.0,
+      "step": 40,
+      "total_flos": 4882092898910208.0,
+      "train_loss": 0.5672883474326227,
+      "train_runtime": 31.3668,
+      "train_samples_per_second": 12.752,
+      "train_steps_per_second": 1.275
     }
   ],
   "logging_steps": 1,
-  "max_steps": 20,
+  "max_steps": 40,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 20,
+  "num_train_epochs": 40,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -334,7 +634,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2482684999761920.0,
+  "total_flos": 4882092898910208.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1252640ae22d0e122a522e37486b3981fc4800f7571ad99afe0f476eb83634a8
+oid sha256:195513e53529c66c5cbfa2e89f4dcaa806198d6a1a878ca5c626aa8f2e024353
 size 6968