Model save

README.md

@@ -1,13 +1,13 @@
 ---
 base_model: meta-llama/Meta-Llama-3-8B
 datasets:
-- llama-duo/synth_summarize_dataset_dedup
+- generator
 library_name: peft
 license: llama3
 tags:
-- alignment-handbook
 - trl
 - sft
+- alignment-handbook
 - generated_from_trainer
 model-index:
 - name: llama3.1-8b-gpt4o_100k_summarize-k
@@ -19,9 +19,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama3.1-8b-gpt4o_100k_summarize-k
 
-This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on the llama-duo/synth_summarize_dataset_dedup dataset.
+This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 2.0198
+- Loss: 2.0191
 
 ## Model description
 
@@ -58,7 +58,7 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.8483        | 1.0   | 195  | 2.0198          |
+| 0.8477        | 1.0   | 195  | 2.0191          |
 
 
 ### Framework versions

all_results.json

@@ -6,9 +6,9 @@
     "eval_samples_per_second": 8.579,
     "eval_steps_per_second": 0.858,
     "total_flos": 5.7571825768464384e+17,
-    "train_loss": 0.9919692577459873,
-    "train_runtime": 2099.4843,
+    "train_loss": 0.9919446505033053,
+    "train_runtime": 2098.6313,
     "train_samples": 115376,
-    "train_samples_per_second": 5.943,
+    "train_samples_per_second": 5.945,
     "train_steps_per_second": 0.093
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 5.7571825768464384e+17,
-    "train_loss": 0.9919692577459873,
-    "train_runtime": 2099.4843,
+    "train_loss": 0.9919446505033053,
+    "train_runtime": 2098.6313,
     "train_samples": 115376,
-    "train_samples_per_second": 5.943,
+    "train_samples_per_second": 5.945,
     "train_steps_per_second": 0.093
 }

trainer_state.json

@@ -10,299 +10,299 @@
   "log_history": [
     {
       "epoch": 0.005128205128205128,
-      "grad_norm": 0.7784392237663269,
+      "grad_norm": 0.7784475088119507,
       "learning_rate": 1.4999999999999999e-05,
       "loss": 2.2555,
       "step": 1
     },
     {
       "epoch": 0.02564102564102564,
-      "grad_norm": 0.9956199526786804,
+      "grad_norm": 0.8670551180839539,
       "learning_rate": 7.5e-05,
-      "loss": 2.2474,
+      "loss": 2.2468,
       "step": 5
     },
     {
       "epoch": 0.05128205128205128,
-      "grad_norm": 1.2538105249404907,
+      "grad_norm": 1.2530401945114136,
       "learning_rate": 0.00015,
-      "loss": 2.1364,
+      "loss": 2.1352,
       "step": 10
     },
     {
       "epoch": 0.07692307692307693,
-      "grad_norm": 1.435035228729248,
+      "grad_norm": 1.460581660270691,
       "learning_rate": 0.000225,
-      "loss": 1.6881,
+      "loss": 1.6865,
       "step": 15
     },
     {
       "epoch": 0.10256410256410256,
-      "grad_norm": 1.9407316446304321,
+      "grad_norm": 2.0667366981506348,
       "learning_rate": 0.0003,
-      "loss": 1.2552,
+      "loss": 1.2589,
       "step": 20
     },
     {
       "epoch": 0.1282051282051282,
-      "grad_norm": 0.6618439555168152,
+      "grad_norm": 0.6875319480895996,
       "learning_rate": 0.00029939614409928584,
-      "loss": 1.1476,
+      "loss": 1.1478,
       "step": 25
     },
     {
       "epoch": 0.15384615384615385,
-      "grad_norm": 0.4018121063709259,
+      "grad_norm": 0.6013126373291016,
       "learning_rate": 0.00029758943828979444,
-      "loss": 1.0741,
+      "loss": 1.076,
       "step": 30
     },
     {
       "epoch": 0.1794871794871795,
-      "grad_norm": 0.3278277814388275,
+      "grad_norm": 0.31481531262397766,
       "learning_rate": 0.00029459442910437797,
-      "loss": 1.0159,
+      "loss": 1.0179,
       "step": 35
     },
     {
       "epoch": 0.20512820512820512,
-      "grad_norm": 0.2918931245803833,
+      "grad_norm": 0.3045659065246582,
       "learning_rate": 0.00029043523059596053,
-      "loss": 0.9818,
+      "loss": 0.9824,
       "step": 40
     },
     {
       "epoch": 0.23076923076923078,
-      "grad_norm": 0.21536950767040253,
+      "grad_norm": 0.2231951653957367,
       "learning_rate": 0.0002851453301853628,
-      "loss": 0.96,
+      "loss": 0.9611,
       "step": 45
     },
     {
       "epoch": 0.2564102564102564,
-      "grad_norm": 0.187772735953331,
+      "grad_norm": 0.18813718855381012,
       "learning_rate": 0.0002787673190402799,
-      "loss": 0.9458,
+      "loss": 0.9461,
       "step": 50
     },
     {
       "epoch": 0.28205128205128205,
-      "grad_norm": 0.2083010971546173,
+      "grad_norm": 0.21799354255199432,
       "learning_rate": 0.0002713525491562421,
-      "loss": 0.9253,
+      "loss": 0.9254,
       "step": 55
     },
     {
       "epoch": 0.3076923076923077,
-      "grad_norm": 0.2108624428510666,
+      "grad_norm": 0.21888460218906403,
       "learning_rate": 0.00026296071990054165,
-      "loss": 0.9218,
+      "loss": 0.9223,
       "step": 60
     },
     {
       "epoch": 0.3333333333333333,
-      "grad_norm": 0.18083609640598297,
+      "grad_norm": 0.18513959646224976,
       "learning_rate": 0.0002536593973480297,
       "loss": 0.9083,
       "step": 65
     },
     {
       "epoch": 0.358974358974359,
-      "grad_norm": 0.178299680352211,
+      "grad_norm": 0.5329850316047668,
       "learning_rate": 0.00024352347027881003,
-      "loss": 0.9036,
+      "loss": 0.9041,
       "step": 70
     },
     {
       "epoch": 0.38461538461538464,
-      "grad_norm": 0.19863168895244598,
+      "grad_norm": 0.19489431381225586,
       "learning_rate": 0.00023263454721781537,
       "loss": 0.9033,
       "step": 75
     },
     {
       "epoch": 0.41025641025641024,
-      "grad_norm": 0.22889919579029083,
+      "grad_norm": 0.2275352030992508,
       "learning_rate": 0.0002210802993709498,
-      "loss": 0.8912,
+      "loss": 0.892,
       "step": 80
     },
     {
       "epoch": 0.4358974358974359,
-      "grad_norm": 0.20287582278251648,
+      "grad_norm": 0.1980154812335968,
       "learning_rate": 0.00020895375474808852,
-      "loss": 0.8899,
+      "loss": 0.8903,
       "step": 85
     },
     {
       "epoch": 0.46153846153846156,
-      "grad_norm": 0.21162137389183044,
+      "grad_norm": 0.20675401389598846,
       "learning_rate": 0.0001963525491562421,
-      "loss": 0.8813,
+      "loss": 0.8815,
       "step": 90
     },
     {
       "epoch": 0.48717948717948717,
-      "grad_norm": 0.21930070221424103,
+      "grad_norm": 0.21656563878059387,
       "learning_rate": 0.00018337814009344714,
-      "loss": 0.8795,
+      "loss": 0.879,
       "step": 95
     },
     {
       "epoch": 0.5128205128205128,
-      "grad_norm": 0.23387882113456726,
+      "grad_norm": 0.22672973573207855,
       "learning_rate": 0.00017013498987264832,
-      "loss": 0.8794,
+      "loss": 0.879,
       "step": 100
     },
     {
       "epoch": 0.5384615384615384,
-      "grad_norm": 0.19868768751621246,
+      "grad_norm": 0.2003437578678131,
       "learning_rate": 0.00015672972455257723,
-      "loss": 0.8752,
+      "loss": 0.8746,
       "step": 105
     },
     {
       "epoch": 0.5641025641025641,
-      "grad_norm": 0.21123158931732178,
+      "grad_norm": 0.21689733862876892,
       "learning_rate": 0.0001432702754474228,
-      "loss": 0.8524,
+      "loss": 0.8522,
       "step": 110
     },
     {
       "epoch": 0.5897435897435898,
-      "grad_norm": 0.25728657841682434,
+      "grad_norm": 0.23330777883529663,
       "learning_rate": 0.0001298650101273517,
-      "loss": 0.8734,
+      "loss": 0.8732,
       "step": 115
     },
     {
       "epoch": 0.6153846153846154,
-      "grad_norm": 0.21154211461544037,
+      "grad_norm": 0.211246058344841,
       "learning_rate": 0.00011662185990655284,
-      "loss": 0.8597,
+      "loss": 0.8594,
       "step": 120
     },
     {
       "epoch": 0.6410256410256411,
-      "grad_norm": 0.21001510322093964,
+      "grad_norm": 0.2110731303691864,
       "learning_rate": 0.0001036474508437579,
-      "loss": 0.8606,
+      "loss": 0.8601,
       "step": 125
     },
     {
       "epoch": 0.6666666666666666,
-      "grad_norm": 0.2222093790769577,
+      "grad_norm": 0.22622446715831757,
       "learning_rate": 9.104624525191145e-05,
-      "loss": 0.8563,
+      "loss": 0.8555,
       "step": 130
     },
     {
       "epoch": 0.6923076923076923,
-      "grad_norm": 0.19596728682518005,
+      "grad_norm": 0.19781175255775452,
       "learning_rate": 7.89197006290502e-05,
-      "loss": 0.855,
+      "loss": 0.8542,
       "step": 135
     },
     {
       "epoch": 0.717948717948718,
-      "grad_norm": 0.20535971224308014,
+      "grad_norm": 0.2107241153717041,
       "learning_rate": 6.736545278218463e-05,
-      "loss": 0.8592,
+      "loss": 0.8587,
       "step": 140
     },
     {
       "epoch": 0.7435897435897436,
-      "grad_norm": 0.20298361778259277,
+      "grad_norm": 0.19920115172863007,
       "learning_rate": 5.6476529721189974e-05,
-      "loss": 0.8549,
+      "loss": 0.8544,
       "step": 145
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 0.19454172253608704,
+      "grad_norm": 0.19098255038261414,
       "learning_rate": 4.63406026519703e-05,
-      "loss": 0.8578,
+      "loss": 0.8572,
       "step": 150
     },
     {
       "epoch": 0.7948717948717948,
-      "grad_norm": 0.19015519320964813,
+      "grad_norm": 0.18926502764225006,
       "learning_rate": 3.7039280099458366e-05,
-      "loss": 0.8579,
+      "loss": 0.8577,
       "step": 155
     },
     {
       "epoch": 0.8205128205128205,
-      "grad_norm": 0.20558574795722961,
+      "grad_norm": 0.20493926107883453,
       "learning_rate": 2.8647450843757897e-05,
-      "loss": 0.8508,
+      "loss": 0.8507,
       "step": 160
     },
     {
       "epoch": 0.8461538461538461,
-      "grad_norm": 0.19159014523029327,
+      "grad_norm": 0.18808531761169434,
       "learning_rate": 2.1232680959720082e-05,
       "loss": 0.8415,
       "step": 165
     },
     {
       "epoch": 0.8717948717948718,
-      "grad_norm": 0.17908993363380432,
+      "grad_norm": 0.18048973381519318,
       "learning_rate": 1.4854669814637143e-05,
-      "loss": 0.8514,
+      "loss": 0.8506,
       "step": 170
     },
     {
       "epoch": 0.8974358974358975,
-      "grad_norm": 0.18227402865886688,
+      "grad_norm": 0.18150238692760468,
       "learning_rate": 9.564769404039419e-06,
-      "loss": 0.8503,
+      "loss": 0.8499,
       "step": 175
     },
     {
       "epoch": 0.9230769230769231,
-      "grad_norm": 0.2028588354587555,
+      "grad_norm": 0.20749987661838531,
       "learning_rate": 5.405570895622013e-06,
-      "loss": 0.8449,
+      "loss": 0.8447,
       "step": 180
     },
     {
       "epoch": 0.9487179487179487,
-      "grad_norm": 0.19046302139759064,
+      "grad_norm": 0.19975171983242035,
       "learning_rate": 2.4105617102055496e-06,
-      "loss": 0.8456,
+      "loss": 0.8446,
       "step": 185
     },
     {
       "epoch": 0.9743589743589743,
-      "grad_norm": 0.18208374083042145,
+      "grad_norm": 0.19153565168380737,
       "learning_rate": 6.038559007141397e-07,
-      "loss": 0.854,
+      "loss": 0.8535,
       "step": 190
     },
     {
       "epoch": 1.0,
-      "grad_norm": 0.21627038717269897,
+      "grad_norm": 0.21618981659412384,
       "learning_rate": 0.0,
-      "loss": 0.8483,
+      "loss": 0.8477,
       "step": 195
     },
     {
       "epoch": 1.0,
-      "eval_loss": 2.0198426246643066,
-      "eval_runtime": 1.151,
-      "eval_samples_per_second": 8.688,
-      "eval_steps_per_second": 0.869,
+      "eval_loss": 2.0191240310668945,
+      "eval_runtime": 1.1347,
+      "eval_samples_per_second": 8.813,
+      "eval_steps_per_second": 0.881,
       "step": 195
     },
     {
       "epoch": 1.0,
       "step": 195,
       "total_flos": 5.7571825768464384e+17,
-      "train_loss": 0.9919692577459873,
-      "train_runtime": 2099.4843,
-      "train_samples_per_second": 5.943,
+      "train_loss": 0.9919446505033053,
+      "train_runtime": 2098.6313,
+      "train_samples_per_second": 5.945,
       "train_steps_per_second": 0.093
     }
   ],