End of training

README.md

@@ -3,10 +3,16 @@ library_name: transformers
 license: apache-2.0
 base_model: alignment-handbook/zephyr-7b-sft-full
 tags:
+- alignment-handbook
+- trl
+- dpo
+- generated_from_trainer
 - trl
 - dpo
 - alignment-handbook
 - generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: zephyr-7b-align-scan
   results: []
@@ -17,7 +23,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-align-scan
 
-This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
+This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the HuggingFaceH4/ultrafeedback_binarized dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.6700
+- Rewards/chosen: 0.0017
+- Rewards/rejected: -0.0506
+- Rewards/accuracies: 0.3155
+- Rewards/margins: 0.0522
+- Logps/rejected: -86.1855
+- Logps/chosen: -74.3248
+- Logits/rejected: -2.4181
+- Logits/chosen: -2.4359
 
 ## Model description
 

all_results.json

@@ -9,10 +9,10 @@
     "eval_rewards/chosen": 0.0016639787936583161,
     "eval_rewards/margins": 0.05223553627729416,
     "eval_rewards/rejected": -0.05057155340909958,
-    "eval_runtime": 113.9808,
+    "eval_runtime": 115.7246,
     "eval_samples": 2000,
-    "eval_samples_per_second": 17.547,
-    "eval_steps_per_second": 0.553,
+    "eval_samples_per_second": 17.282,
+    "eval_steps_per_second": 0.544,
     "total_flos": 0.0,
     "train_loss": 0.0,
     "train_runtime": 0.0191,

config.json

@@ -22,6 +22,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -9,8 +9,8 @@
     "eval_rewards/chosen": 0.0016639787936583161,
     "eval_rewards/margins": 0.05223553627729416,
     "eval_rewards/rejected": -0.05057155340909958,
-    "eval_runtime": 113.9808,
+    "eval_runtime": 115.7246,
     "eval_samples": 2000,
-    "eval_samples_per_second": 17.547,
-    "eval_steps_per_second": 0.553
+    "eval_samples_per_second": 17.282,
+    "eval_steps_per_second": 0.544
 }