End of training

README.md

@@ -2,9 +2,15 @@
 license: apache-2.0
 base_model: Minbyul/mistral-7b-wo-healthsearch_qa-sft
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: mistral-7b-dpo-full-sft-wo-healthsearch_qa
   results: []
@@ -15,7 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # mistral-7b-dpo-full-sft-wo-healthsearch_qa
 
-This model is a fine-tuned version of [Minbyul/mistral-7b-wo-healthsearch_qa-sft](https://huggingface.co/Minbyul/mistral-7b-wo-healthsearch_qa-sft) on an unknown dataset.
+This model is a fine-tuned version of [Minbyul/mistral-7b-wo-healthsearch_qa-sft](https://huggingface.co/Minbyul/mistral-7b-wo-healthsearch_qa-sft) on the HuggingFaceH4/ultrafeedback_binarized dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.6746
+- Rewards/chosen: -0.0204
+- Rewards/rejected: -0.0600
+- Rewards/accuracies: 0.6612
+- Rewards/margins: 0.0395
+- Logps/rejected: -1091.8407
+- Logps/chosen: -817.4551
+- Logits/rejected: -2.8353
+- Logits/chosen: -2.9083
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.9083194732666016,
+    "eval_logits/rejected": -2.835282325744629,
+    "eval_logps/chosen": -817.455078125,
+    "eval_logps/rejected": -1091.8406982421875,
+    "eval_loss": 0.6746304631233215,
+    "eval_rewards/accuracies": 0.6612318754196167,
+    "eval_rewards/chosen": -0.020445356145501137,
+    "eval_rewards/margins": 0.039513371884822845,
+    "eval_rewards/rejected": -0.059958722442388535,
+    "eval_runtime": 2163.8157,
+    "eval_samples": 19864,
+    "eval_samples_per_second": 9.18,
+    "eval_steps_per_second": 0.287,
     "train_loss": 0.5004948377609253,
     "train_runtime": 257.4676,
     "train_samples": 1128,

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -2.9083194732666016,
+    "eval_logits/rejected": -2.835282325744629,
+    "eval_logps/chosen": -817.455078125,
+    "eval_logps/rejected": -1091.8406982421875,
+    "eval_loss": 0.6746304631233215,
+    "eval_rewards/accuracies": 0.6612318754196167,
+    "eval_rewards/chosen": -0.020445356145501137,
+    "eval_rewards/margins": 0.039513371884822845,
+    "eval_rewards/rejected": -0.059958722442388535,
+    "eval_runtime": 2163.8157,
+    "eval_samples": 19864,
+    "eval_samples_per_second": 9.18,
+    "eval_steps_per_second": 0.287
+}