End of training

README.md

@@ -3,9 +3,15 @@ library_name: transformers
 license: apache-2.0
 base_model: mistralai/Mistral-7B-Instruct-v0.2
 tags:
+- alignment-handbook
 - trl
 - simpo
 - generated_from_trainer
+- trl
+- simpo
+- generated_from_trainer
+datasets:
+- princeton-nlp/mistral-instruct-ultrafeedback
 model-index:
 - name: mistral-7b-instruct-v0.2-rsimpo-full
   results: []
@@ -16,7 +22,16 @@ should probably proofread and complete it, then remove this comment. -->
 
 # mistral-7b-instruct-v0.2-rsimpo-full
 
-This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on an unknown dataset.
+This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the princeton-nlp/mistral-instruct-ultrafeedback dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.7976
+- Rewards/chosen: -1.2291
+- Rewards/rejected: -1.3244
+- Rewards/accuracies: 0.5507
+- Rewards/margins: 0.0955
+- Agreement Weights/mean: 0.9696
+- Agreement Weights/std: 0.0182
+- Eta/annotator 0: 0.9730
 
 ## Model description
 

all_results.json

@@ -1,5 +1,17 @@
 {
     "epoch": 0.998660594695955,
+    "eta/annotator_0": 0.9730426073074341,
+    "eval_agreement_weights/mean": 0.9695923328399658,
+    "eval_agreement_weights/std": 0.018151868134737015,
+    "eval_loss": 0.7976000905036926,
+    "eval_rewards/accuracies": 0.5507165789604187,
+    "eval_rewards/chosen": -1.2290558815002441,
+    "eval_rewards/margins": 0.09547457098960876,
+    "eval_rewards/rejected": -1.3243849277496338,
+    "eval_runtime": 101.569,
+    "eval_samples": 2994,
+    "eval_samples_per_second": 29.478,
+    "eval_steps_per_second": 0.925,
     "total_flos": 0.0,
     "train_loss": 0.8001983071089814,
     "train_runtime": 7557.1271,

config.json

@@ -22,6 +22,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.46.3",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 0.998660594695955,
-    "eta/annotator_0": 0.9783164262771606,
-    "eval_agreement_weights/mean": 0.976341724395752,
-    "eval_agreement_weights/std": 0.0132759353145957,
-    "eval_loss": 0.8032518625259399,
-    "eval_rewards/accuracies": 0.554964542388916,
-    "eval_rewards/chosen": -1.2927193641662598,
-    "eval_rewards/margins": 0.08169101178646088,
-    "eval_rewards/rejected": -1.374501347541809,
-    "eval_runtime": 63.5366,
+    "eta/annotator_0": 0.9730426073074341,
+    "eval_agreement_weights/mean": 0.9695923328399658,
+    "eval_agreement_weights/std": 0.018151868134737015,
+    "eval_loss": 0.7976000905036926,
+    "eval_rewards/accuracies": 0.5507165789604187,
+    "eval_rewards/chosen": -1.2290558815002441,
+    "eval_rewards/margins": 0.09547457098960876,
+    "eval_rewards/rejected": -1.3243849277496338,
+    "eval_runtime": 101.569,
     "eval_samples": 2994,
-    "eval_samples_per_second": 47.122,
-    "eval_steps_per_second": 1.479
+    "eval_samples_per_second": 29.478,
+    "eval_steps_per_second": 0.925
 }