Model save

README.md

@@ -4,7 +4,6 @@ library_name: transformers
 model_name: llama_check_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/pq3g1op3)
+[<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/ierr3rjg)
 
 This model was trained with SFT.
 

all_results.json

@@ -5,10 +5,10 @@
     "eval_samples": 10,
     "eval_samples_per_second": 89.351,
     "eval_steps_per_second": 17.87,
-    "total_flos": 3352730995261440.0,
-    "train_loss": 0.9132046818675008,
-    "train_runtime": 21.9272,
+    "total_flos": 4849620932886528.0,
+    "train_loss": 0.6959351973579032,
+    "train_runtime": 31.7849,
     "train_samples": 10,
-    "train_samples_per_second": 18.242,
-    "train_steps_per_second": 1.824
+    "train_samples_per_second": 12.585,
+    "train_steps_per_second": 1.258
 }

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:afb8bc60035e6f5c6b9aa9bc817024448f290884b8767593e9aecb1baf5f61bb
+oid sha256:b1c5a4c3b8aa8f79f141452fcaf4d59840224c3af6c6b366b00f97903fe6c4e0
 size 4965799096

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:364d194401bb4084b3ee664086eefd10bb5df26de57475f6344b62780ee2e7d9
+oid sha256:dba518352df14b4438c4b2b16ddd67fa6fb811472fd5fb67395889f198be0c71
 size 2247734992

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6fca926fdfedcb8fd225765fb3911e0ab5de0390ec7242a46dff829e8604f723
-size 17210020
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 40.0,
-    "total_flos": 3352730995261440.0,
-    "train_loss": 0.9132046818675008,
-    "train_runtime": 21.9272,
+    "total_flos": 4849620932886528.0,
+    "train_loss": 0.6959351973579032,
+    "train_runtime": 31.7849,
     "train_samples": 10,
-    "train_samples_per_second": 18.242,
-    "train_steps_per_second": 1.824
+    "train_samples_per_second": 12.585,
+    "train_steps_per_second": 1.258
 }

trainer_state.json

@@ -11,610 +11,610 @@
     {
       "epoch": 1.0,
       "learning_rate": 2.5e-05,
-      "loss": 1.1193,
+      "loss": 1.1447,
       "step": 1
     },
     {
       "epoch": 1.0,
-      "eval_loss": 1.1183021068572998,
-      "eval_runtime": 0.097,
-      "eval_samples_per_second": 103.081,
-      "eval_steps_per_second": 20.616,
+      "eval_loss": 1.1312110424041748,
+      "eval_runtime": 0.1567,
+      "eval_samples_per_second": 63.798,
+      "eval_steps_per_second": 12.76,
       "step": 1
     },
     {
       "epoch": 2.0,
-      "grad_norm": 19.4419002532959,
+      "grad_norm": 9.785662651062012,
       "learning_rate": 5e-05,
-      "loss": 1.1193,
+      "loss": 1.1235,
       "step": 2
     },
     {
       "epoch": 2.0,
-      "eval_loss": 0.4163738191127777,
-      "eval_runtime": 0.0967,
-      "eval_samples_per_second": 103.452,
-      "eval_steps_per_second": 20.69,
+      "eval_loss": 0.7337616682052612,
+      "eval_runtime": 0.156,
+      "eval_samples_per_second": 64.086,
+      "eval_steps_per_second": 12.817,
       "step": 2
     },
     {
       "epoch": 3.0,
-      "grad_norm": 19.4419002532959,
+      "grad_norm": 9.785662651062012,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 0.4152,
+      "loss": 0.743,
       "step": 3
     },
     {
       "epoch": 3.0,
-      "eval_loss": 0.4163738191127777,
-      "eval_runtime": 0.0975,
-      "eval_samples_per_second": 102.538,
-      "eval_steps_per_second": 20.508,
+      "eval_loss": 0.7337616682052612,
+      "eval_runtime": 0.1556,
+      "eval_samples_per_second": 64.283,
+      "eval_steps_per_second": 12.857,
       "step": 3
     },
     {
       "epoch": 4.0,
-      "grad_norm": 12.695544242858887,
+      "grad_norm": 6.636180877685547,
       "learning_rate": 0.0001,
-      "loss": 0.4152,
+      "loss": 0.7504,
       "step": 4
     },
     {
       "epoch": 4.0,
-      "eval_loss": 0.8865498304367065,
-      "eval_runtime": 0.0966,
-      "eval_samples_per_second": 103.475,
-      "eval_steps_per_second": 20.695,
+      "eval_loss": 1.1203209161758423,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.304,
+      "eval_steps_per_second": 12.861,
       "step": 4
     },
     {
       "epoch": 5.0,
-      "grad_norm": 12.695544242858887,
+      "grad_norm": 6.636180877685547,
       "learning_rate": 0.000125,
-      "loss": 0.8864,
+      "loss": 1.1145,
       "step": 5
     },
     {
       "epoch": 5.0,
-      "eval_loss": 0.8865498304367065,
-      "eval_runtime": 0.0977,
-      "eval_samples_per_second": 102.389,
-      "eval_steps_per_second": 20.478,
+      "eval_loss": 1.1203209161758423,
+      "eval_runtime": 0.1562,
+      "eval_samples_per_second": 64.004,
+      "eval_steps_per_second": 12.801,
       "step": 5
     },
     {
       "epoch": 6.0,
-      "grad_norm": 39.317420959472656,
+      "grad_norm": 18.654922485351562,
       "learning_rate": 0.00015000000000000001,
-      "loss": 0.8864,
+      "loss": 1.0779,
       "step": 6
     },
     {
       "epoch": 6.0,
-      "eval_loss": 1.6666438579559326,
-      "eval_runtime": 0.0998,
-      "eval_samples_per_second": 100.229,
-      "eval_steps_per_second": 20.046,
+      "eval_loss": 0.9797636866569519,
+      "eval_runtime": 0.1554,
+      "eval_samples_per_second": 64.341,
+      "eval_steps_per_second": 12.868,
       "step": 6
     },
     {
       "epoch": 7.0,
-      "grad_norm": 39.317420959472656,
+      "grad_norm": 18.654922485351562,
       "learning_rate": 0.000175,
-      "loss": 1.6618,
+      "loss": 0.9815,
       "step": 7
     },
     {
       "epoch": 7.0,
-      "eval_loss": 1.6666438579559326,
-      "eval_runtime": 0.0975,
-      "eval_samples_per_second": 102.551,
-      "eval_steps_per_second": 20.51,
+      "eval_loss": 0.9797636866569519,
+      "eval_runtime": 0.1551,
+      "eval_samples_per_second": 64.482,
+      "eval_steps_per_second": 12.896,
       "step": 7
     },
     {
       "epoch": 8.0,
-      "grad_norm": 44.225120544433594,
+      "grad_norm": 13.476932525634766,
       "learning_rate": 0.0002,
-      "loss": 1.6618,
+      "loss": 1.0471,
       "step": 8
     },
     {
       "epoch": 8.0,
-      "eval_loss": 3.048259973526001,
-      "eval_runtime": 0.0983,
-      "eval_samples_per_second": 101.777,
-      "eval_steps_per_second": 20.355,
+      "eval_loss": 5.148657321929932,
+      "eval_runtime": 0.156,
+      "eval_samples_per_second": 64.108,
+      "eval_steps_per_second": 12.822,
       "step": 8
     },
     {
       "epoch": 9.0,
-      "grad_norm": 44.225120544433594,
+      "grad_norm": 13.476932525634766,
       "learning_rate": 0.0001995184726672197,
-      "loss": 3.0505,
+      "loss": 5.1291,
       "step": 9
     },
     {
       "epoch": 9.0,
-      "eval_loss": 3.048259973526001,
-      "eval_runtime": 0.0993,
-      "eval_samples_per_second": 100.727,
-      "eval_steps_per_second": 20.145,
+      "eval_loss": 5.148657321929932,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.323,
+      "eval_steps_per_second": 12.865,
       "step": 9
     },
     {
       "epoch": 10.0,
-      "grad_norm": 143.0501251220703,
+      "grad_norm": 141.67774963378906,
       "learning_rate": 0.00019807852804032305,
-      "loss": 3.0505,
+      "loss": 5.2219,
       "step": 10
     },
     {
       "epoch": 10.0,
-      "eval_loss": 2.2029616832733154,
-      "eval_runtime": 0.097,
-      "eval_samples_per_second": 103.098,
-      "eval_steps_per_second": 20.62,
+      "eval_loss": 2.3491921424865723,
+      "eval_runtime": 0.1552,
+      "eval_samples_per_second": 64.421,
+      "eval_steps_per_second": 12.884,
       "step": 10
     },
     {
       "epoch": 11.0,
-      "grad_norm": 143.0501251220703,
+      "grad_norm": 141.67774963378906,
       "learning_rate": 0.0001956940335732209,
-      "loss": 2.2026,
+      "loss": 2.4209,
       "step": 11
     },
     {
       "epoch": 11.0,
-      "eval_loss": 2.2029616832733154,
-      "eval_runtime": 0.0976,
-      "eval_samples_per_second": 102.442,
-      "eval_steps_per_second": 20.488,
+      "eval_loss": 2.3491921424865723,
+      "eval_runtime": 0.156,
+      "eval_samples_per_second": 64.119,
+      "eval_steps_per_second": 12.824,
       "step": 11
     },
     {
       "epoch": 12.0,
-      "grad_norm": 27.46721076965332,
+      "grad_norm": 95.97136688232422,
       "learning_rate": 0.0001923879532511287,
-      "loss": 2.2026,
+      "loss": 2.4686,
       "step": 12
     },
     {
       "epoch": 12.0,
-      "eval_loss": 2.8971307277679443,
-      "eval_runtime": 0.0976,
-      "eval_samples_per_second": 102.486,
-      "eval_steps_per_second": 20.497,
+      "eval_loss": 0.9826704859733582,
+      "eval_runtime": 0.1558,
+      "eval_samples_per_second": 64.17,
+      "eval_steps_per_second": 12.834,
       "step": 12
     },
     {
       "epoch": 13.0,
-      "grad_norm": 27.46721076965332,
+      "grad_norm": 95.97136688232422,
       "learning_rate": 0.0001881921264348355,
-      "loss": 2.9002,
+      "loss": 0.986,
       "step": 13
     },
     {
       "epoch": 13.0,
-      "eval_loss": 2.8971307277679443,
-      "eval_runtime": 0.0973,
-      "eval_samples_per_second": 102.809,
-      "eval_steps_per_second": 20.562,
+      "eval_loss": 0.9826704859733582,
+      "eval_runtime": 0.1552,
+      "eval_samples_per_second": 64.426,
+      "eval_steps_per_second": 12.885,
       "step": 13
     },
     {
       "epoch": 14.0,
-      "grad_norm": 49.42749786376953,
+      "grad_norm": 7.2887864112854,
       "learning_rate": 0.00018314696123025454,
-      "loss": 2.9002,
+      "loss": 0.9912,
       "step": 14
     },
     {
       "epoch": 14.0,
-      "eval_loss": 3.4301562309265137,
-      "eval_runtime": 0.0984,
-      "eval_samples_per_second": 101.612,
-      "eval_steps_per_second": 20.322,
+      "eval_loss": 0.7040748596191406,
+      "eval_runtime": 0.1558,
+      "eval_samples_per_second": 64.185,
+      "eval_steps_per_second": 12.837,
       "step": 14
     },
     {
       "epoch": 15.0,
-      "grad_norm": 49.42749786376953,
+      "grad_norm": 7.2887864112854,
       "learning_rate": 0.0001773010453362737,
-      "loss": 3.4308,
+      "loss": 0.7151,
       "step": 15
     },
     {
       "epoch": 15.0,
-      "eval_loss": 3.4301562309265137,
-      "eval_runtime": 0.0981,
-      "eval_samples_per_second": 101.972,
-      "eval_steps_per_second": 20.394,
+      "eval_loss": 0.7040748596191406,
+      "eval_runtime": 0.1556,
+      "eval_samples_per_second": 64.283,
+      "eval_steps_per_second": 12.857,
       "step": 15
     },
     {
       "epoch": 16.0,
-      "grad_norm": 46.989479064941406,
+      "grad_norm": 7.857557773590088,
       "learning_rate": 0.00017071067811865476,
-      "loss": 3.4308,
+      "loss": 0.7103,
       "step": 16
     },
     {
       "epoch": 16.0,
-      "eval_loss": 1.4099053144454956,
-      "eval_runtime": 0.0973,
-      "eval_samples_per_second": 102.796,
-      "eval_steps_per_second": 20.559,
+      "eval_loss": 0.27107304334640503,
+      "eval_runtime": 0.1552,
+      "eval_samples_per_second": 64.413,
+      "eval_steps_per_second": 12.883,
       "step": 16
     },
     {
       "epoch": 17.0,
-      "grad_norm": 46.989479064941406,
+      "grad_norm": 7.857557773590088,
       "learning_rate": 0.00016343932841636456,
-      "loss": 1.4133,
+      "loss": 0.2735,
       "step": 17
     },
     {
       "epoch": 17.0,
-      "eval_loss": 1.4099053144454956,
-      "eval_runtime": 0.097,
-      "eval_samples_per_second": 103.061,
-      "eval_steps_per_second": 20.612,
+      "eval_loss": 0.27107304334640503,
+      "eval_runtime": 0.1552,
+      "eval_samples_per_second": 64.433,
+      "eval_steps_per_second": 12.887,
       "step": 17
     },
     {
       "epoch": 18.0,
-      "grad_norm": 19.974260330200195,
+      "grad_norm": 3.201599597930908,
       "learning_rate": 0.00015555702330196023,
-      "loss": 1.4133,
+      "loss": 0.2697,
       "step": 18
     },
     {
       "epoch": 18.0,
-      "eval_loss": 0.7468957304954529,
-      "eval_runtime": 0.0986,
-      "eval_samples_per_second": 101.448,
-      "eval_steps_per_second": 20.29,
+      "eval_loss": 0.15566985309123993,
+      "eval_runtime": 0.1556,
+      "eval_samples_per_second": 64.252,
+      "eval_steps_per_second": 12.85,
       "step": 18
     },
     {
       "epoch": 19.0,
-      "grad_norm": 19.974260330200195,
+      "grad_norm": 3.201599597930908,
       "learning_rate": 0.0001471396736825998,
-      "loss": 0.746,
+      "loss": 0.1503,
       "step": 19
     },
     {
       "epoch": 19.0,
-      "eval_loss": 0.7468957304954529,
-      "eval_runtime": 0.0972,
-      "eval_samples_per_second": 102.877,
-      "eval_steps_per_second": 20.575,
+      "eval_loss": 0.15566985309123993,
+      "eval_runtime": 0.1647,
+      "eval_samples_per_second": 60.711,
+      "eval_steps_per_second": 12.142,
       "step": 19
     },
     {
       "epoch": 20.0,
-      "grad_norm": 9.610711097717285,
+      "grad_norm": 2.2040343284606934,
       "learning_rate": 0.000138268343236509,
-      "loss": 0.746,
+      "loss": 0.1456,
       "step": 20
     },
     {
       "epoch": 20.0,
-      "eval_loss": 0.16501614451408386,
-      "eval_runtime": 0.1002,
-      "eval_samples_per_second": 99.82,
-      "eval_steps_per_second": 19.964,
+      "eval_loss": 0.10551667213439941,
+      "eval_runtime": 0.1561,
+      "eval_samples_per_second": 64.064,
+      "eval_steps_per_second": 12.813,
       "step": 20
     },
     {
       "epoch": 21.0,
-      "grad_norm": 9.610711097717285,
+      "grad_norm": 2.2040343284606934,
       "learning_rate": 0.00012902846772544624,
-      "loss": 0.1654,
+      "loss": 0.1046,
       "step": 21
     },
     {
       "epoch": 21.0,
-      "eval_loss": 0.16501614451408386,
-      "eval_runtime": 0.0976,
-      "eval_samples_per_second": 102.477,
-      "eval_steps_per_second": 20.495,
+      "eval_loss": 0.10551667213439941,
+      "eval_runtime": 0.1585,
+      "eval_samples_per_second": 63.083,
+      "eval_steps_per_second": 12.617,
       "step": 21
     },
     {
       "epoch": 22.0,
-      "grad_norm": 3.4515697956085205,
+      "grad_norm": 1.8842490911483765,
       "learning_rate": 0.00011950903220161285,
-      "loss": 0.1654,
+      "loss": 0.1046,
       "step": 22
     },
     {
       "epoch": 22.0,
-      "eval_loss": 0.14767812192440033,
-      "eval_runtime": 0.097,
-      "eval_samples_per_second": 103.11,
-      "eval_steps_per_second": 20.622,
+      "eval_loss": 0.03636971116065979,
+      "eval_runtime": 0.1553,
+      "eval_samples_per_second": 64.41,
+      "eval_steps_per_second": 12.882,
       "step": 22
     },
     {
       "epoch": 23.0,
-      "grad_norm": 3.4515697956085205,
+      "grad_norm": 1.8842490911483765,
       "learning_rate": 0.0001098017140329561,
-      "loss": 0.1474,
+      "loss": 0.0369,
       "step": 23
     },
     {
       "epoch": 23.0,
-      "eval_loss": 0.14767812192440033,
-      "eval_runtime": 0.1041,
-      "eval_samples_per_second": 96.05,
-      "eval_steps_per_second": 19.21,
+      "eval_loss": 0.03636971116065979,
+      "eval_runtime": 0.1564,
+      "eval_samples_per_second": 63.924,
+      "eval_steps_per_second": 12.785,
       "step": 23
     },
     {
       "epoch": 24.0,
-      "grad_norm": 4.262996196746826,
+      "grad_norm": 0.7897646427154541,
       "learning_rate": 0.0001,
-      "loss": 0.1474,
+      "loss": 0.0332,
       "step": 24
     },
     {
       "epoch": 24.0,
-      "eval_loss": 0.05527349188923836,
-      "eval_runtime": 0.0983,
-      "eval_samples_per_second": 101.742,
-      "eval_steps_per_second": 20.348,
+      "eval_loss": 0.021218404173851013,
+      "eval_runtime": 0.1562,
+      "eval_samples_per_second": 64.028,
+      "eval_steps_per_second": 12.806,
       "step": 24
     },
     {
       "epoch": 25.0,
-      "grad_norm": 4.262996196746826,
+      "grad_norm": 0.7897646427154541,
       "learning_rate": 9.019828596704394e-05,
-      "loss": 0.0569,
+      "loss": 0.0229,
       "step": 25
     },
     {
       "epoch": 25.0,
-      "eval_loss": 0.05527349188923836,
-      "eval_runtime": 0.1002,
-      "eval_samples_per_second": 99.795,
-      "eval_steps_per_second": 19.959,
+      "eval_loss": 0.021218404173851013,
+      "eval_runtime": 0.1565,
+      "eval_samples_per_second": 63.916,
+      "eval_steps_per_second": 12.783,
       "step": 25
     },
     {
       "epoch": 26.0,
-      "grad_norm": 2.504621744155884,
+      "grad_norm": 0.553653359413147,
       "learning_rate": 8.049096779838719e-05,
-      "loss": 0.0569,
+      "loss": 0.0188,
       "step": 26
     },
     {
       "epoch": 26.0,
-      "eval_loss": 0.05637549236416817,
-      "eval_runtime": 0.1041,
-      "eval_samples_per_second": 96.026,
-      "eval_steps_per_second": 19.205,
+      "eval_loss": 0.010966768488287926,
+      "eval_runtime": 0.1552,
+      "eval_samples_per_second": 64.414,
+      "eval_steps_per_second": 12.883,
       "step": 26
     },
     {
       "epoch": 27.0,
-      "grad_norm": 2.504621744155884,
+      "grad_norm": 0.553653359413147,
       "learning_rate": 7.097153227455379e-05,
-      "loss": 0.0575,
+      "loss": 0.0096,
       "step": 27
     },
     {
       "epoch": 27.0,
-      "eval_loss": 0.05637549236416817,
-      "eval_runtime": 0.0973,
-      "eval_samples_per_second": 102.725,
-      "eval_steps_per_second": 20.545,
+      "eval_loss": 0.010966768488287926,
+      "eval_runtime": 0.1573,
+      "eval_samples_per_second": 63.588,
+      "eval_steps_per_second": 12.718,
       "step": 27
     },
     {
       "epoch": 28.0,
-      "grad_norm": 3.5743298530578613,
+      "grad_norm": 0.2734664976596832,
       "learning_rate": 6.173165676349103e-05,
-      "loss": 0.0575,
+      "loss": 0.0085,
       "step": 28
     },
     {
       "epoch": 28.0,
-      "eval_loss": 0.003984498791396618,
-      "eval_runtime": 0.0975,
-      "eval_samples_per_second": 102.605,
-      "eval_steps_per_second": 20.521,
+      "eval_loss": 0.010634347796440125,
+      "eval_runtime": 0.1562,
+      "eval_samples_per_second": 64.024,
+      "eval_steps_per_second": 12.805,
       "step": 28
     },
     {
       "epoch": 29.0,
-      "grad_norm": 3.5743298530578613,
+      "grad_norm": 0.2734664976596832,
       "learning_rate": 5.286032631740023e-05,
-      "loss": 0.0041,
+      "loss": 0.0092,
       "step": 29
     },
     {
       "epoch": 29.0,
-      "eval_loss": 0.003984498791396618,
-      "eval_runtime": 0.0979,
-      "eval_samples_per_second": 102.094,
-      "eval_steps_per_second": 20.419,
+      "eval_loss": 0.010634347796440125,
+      "eval_runtime": 0.1573,
+      "eval_samples_per_second": 63.573,
+      "eval_steps_per_second": 12.715,
       "step": 29
     },
     {
       "epoch": 30.0,
-      "grad_norm": 0.1635422706604004,
+      "grad_norm": 0.8116740584373474,
       "learning_rate": 4.444297669803981e-05,
-      "loss": 0.0041,
+      "loss": 0.0086,
       "step": 30
     },
     {
       "epoch": 30.0,
-      "eval_loss": 0.0021899878047406673,
-      "eval_runtime": 0.098,
-      "eval_samples_per_second": 102.061,
-      "eval_steps_per_second": 20.412,
+      "eval_loss": 0.002800833899527788,
+      "eval_runtime": 0.1563,
+      "eval_samples_per_second": 63.985,
+      "eval_steps_per_second": 12.797,
       "step": 30
     },
     {
       "epoch": 31.0,
-      "grad_norm": 0.1635422706604004,
+      "grad_norm": 0.8116740584373474,
       "learning_rate": 3.6560671583635467e-05,
-      "loss": 0.0022,
+      "loss": 0.0023,
       "step": 31
     },
     {
       "epoch": 31.0,
-      "eval_loss": 0.0021899878047406673,
-      "eval_runtime": 0.0971,
-      "eval_samples_per_second": 103.036,
-      "eval_steps_per_second": 20.607,
+      "eval_loss": 0.002800833899527788,
+      "eval_runtime": 0.1557,
+      "eval_samples_per_second": 64.242,
+      "eval_steps_per_second": 12.848,
       "step": 31
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.0847228392958641,
+      "grad_norm": 0.13856175541877747,
       "learning_rate": 2.9289321881345254e-05,
-      "loss": 0.0022,
+      "loss": 0.0028,
       "step": 32
     },
     {
       "epoch": 32.0,
-      "eval_loss": 0.0015291140880435705,
-      "eval_runtime": 0.1006,
-      "eval_samples_per_second": 99.357,
-      "eval_steps_per_second": 19.871,
+      "eval_loss": 0.0017185775795951486,
+      "eval_runtime": 0.1572,
+      "eval_samples_per_second": 63.629,
+      "eval_steps_per_second": 12.726,
       "step": 32
     },
     {
       "epoch": 33.0,
-      "grad_norm": 0.0847228392958641,
+      "grad_norm": 0.13856175541877747,
       "learning_rate": 2.26989546637263e-05,
-      "loss": 0.0015,
+      "loss": 0.0016,
       "step": 33
     },
     {
       "epoch": 33.0,
-      "eval_loss": 0.0015291140880435705,
-      "eval_runtime": 0.1027,
-      "eval_samples_per_second": 97.387,
-      "eval_steps_per_second": 19.477,
+      "eval_loss": 0.0017185775795951486,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.322,
+      "eval_steps_per_second": 12.864,
       "step": 33
     },
     {
       "epoch": 34.0,
-      "grad_norm": 0.06535077840089798,
+      "grad_norm": 0.050298936665058136,
       "learning_rate": 1.6853038769745467e-05,
-      "loss": 0.0015,
+      "loss": 0.0016,
       "step": 34
     },
     {
       "epoch": 34.0,
-      "eval_loss": 0.0011607444612309337,
-      "eval_runtime": 0.0977,
-      "eval_samples_per_second": 102.397,
-      "eval_steps_per_second": 20.479,
+      "eval_loss": 0.0014206025516614318,
+      "eval_runtime": 0.1555,
+      "eval_samples_per_second": 64.317,
+      "eval_steps_per_second": 12.863,
       "step": 34
     },
     {
       "epoch": 35.0,
-      "grad_norm": 0.06535077840089798,
+      "grad_norm": 0.050298936665058136,
       "learning_rate": 1.1807873565164506e-05,
-      "loss": 0.0011,
+      "loss": 0.0013,
       "step": 35
     },
     {
       "epoch": 35.0,
-      "eval_loss": 0.0011607444612309337,
-      "eval_runtime": 0.0971,
-      "eval_samples_per_second": 102.985,
-      "eval_steps_per_second": 20.597,
+      "eval_loss": 0.0014206025516614318,
+      "eval_runtime": 0.1566,
+      "eval_samples_per_second": 63.862,
+      "eval_steps_per_second": 12.772,
       "step": 35
     },
     {
       "epoch": 36.0,
-      "grad_norm": 0.04596730321645737,
+      "grad_norm": 0.04518803581595421,
       "learning_rate": 7.612046748871327e-06,
-      "loss": 0.0011,
+      "loss": 0.0013,
       "step": 36
     },
     {
       "epoch": 36.0,
-      "eval_loss": 0.0010076254839077592,
-      "eval_runtime": 0.0972,
-      "eval_samples_per_second": 102.913,
-      "eval_steps_per_second": 20.583,
+      "eval_loss": 0.0012987955706194043,
+      "eval_runtime": 0.1581,
+      "eval_samples_per_second": 63.253,
+      "eval_steps_per_second": 12.651,
       "step": 36
     },
     {
       "epoch": 37.0,
-      "grad_norm": 0.04596730321645737,
+      "grad_norm": 0.04518803581595421,
       "learning_rate": 4.305966426779118e-06,
-      "loss": 0.001,
+      "loss": 0.0013,
       "step": 37
     },
     {
       "epoch": 37.0,
-      "eval_loss": 0.0010076254839077592,
-      "eval_runtime": 0.097,
-      "eval_samples_per_second": 103.097,
-      "eval_steps_per_second": 20.619,
+      "eval_loss": 0.0012987955706194043,
+      "eval_runtime": 0.1552,
+      "eval_samples_per_second": 64.423,
+      "eval_steps_per_second": 12.885,
       "step": 37
     },
     {
       "epoch": 38.0,
-      "grad_norm": 0.039849672466516495,
+      "grad_norm": 0.04950498044490814,
       "learning_rate": 1.921471959676957e-06,
-      "loss": 0.001,
+      "loss": 0.0012,
       "step": 38
     },
     {
       "epoch": 38.0,
-      "eval_loss": 0.0009246615809388459,
-      "eval_runtime": 0.0975,
-      "eval_samples_per_second": 102.584,
-      "eval_steps_per_second": 20.517,
+      "eval_loss": 0.0012480242876335979,
+      "eval_runtime": 0.1557,
+      "eval_samples_per_second": 64.216,
+      "eval_steps_per_second": 12.843,
       "step": 38
     },
     {
       "epoch": 39.0,
-      "grad_norm": 0.039849672466516495,
+      "grad_norm": 0.04950498044490814,
       "learning_rate": 4.815273327803182e-07,
-      "loss": 0.0009,
+      "loss": 0.0012,
       "step": 39
     },
     {
       "epoch": 39.0,
-      "eval_loss": 0.0009246615809388459,
-      "eval_runtime": 0.099,
-      "eval_samples_per_second": 101.052,
-      "eval_steps_per_second": 20.21,
+      "eval_loss": 0.0012480242876335979,
+      "eval_runtime": 0.1593,
+      "eval_samples_per_second": 62.756,
+      "eval_steps_per_second": 12.551,
       "step": 39
     },
     {
       "epoch": 40.0,
-      "grad_norm": 0.037203624844551086,
+      "grad_norm": 0.05591588839888573,
       "learning_rate": 0.0,
-      "loss": 0.0009,
+      "loss": 0.0011,
       "step": 40
     },
     {
       "epoch": 40.0,
-      "eval_loss": 0.0009156091255135834,
-      "eval_runtime": 0.0977,
-      "eval_samples_per_second": 102.33,
-      "eval_steps_per_second": 20.466,
+      "eval_loss": 0.0012466020416468382,
+      "eval_runtime": 0.1587,
+      "eval_samples_per_second": 63.027,
+      "eval_steps_per_second": 12.605,
       "step": 40
     },
     {
       "epoch": 40.0,
       "step": 40,
-      "total_flos": 3352730995261440.0,
-      "train_loss": 0.9132046818675008,
-      "train_runtime": 21.9272,
-      "train_samples_per_second": 18.242,
-      "train_steps_per_second": 1.824
+      "total_flos": 4849620932886528.0,
+      "train_loss": 0.6959351973579032,
+      "train_runtime": 31.7849,
+      "train_samples_per_second": 12.585,
+      "train_steps_per_second": 1.258
     }
   ],
   "logging_steps": 1,
@@ -634,7 +634,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3352730995261440.0,
+  "total_flos": 4849620932886528.0,
   "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:fa2c887c5cce835e5cbf3d2e0b70db81040dda2b3369db7f54f3b1f24d181091
+oid sha256:ba9b06b8bcd73c20d1ab29d48efa07a9c52ce282d69ea2221de1b9ccde6ccb65
 size 6968