End of training

README.md

@@ -1,10 +1,16 @@
 ---
 base_model: Minbyul/selfbiorag-7b-wo-medication_qa-sft
 tags:
+- alignment-handbook
+- trl
+- dpo
+- generated_from_trainer
 - trl
 - dpo
 - alignment-handbook
 - generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: selfbiorag-7b-dpo-full-sft-wo-medication_qa
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # selfbiorag-7b-dpo-full-sft-wo-medication_qa
 
-This model is a fine-tuned version of [Minbyul/selfbiorag-7b-wo-medication_qa-sft](https://huggingface.co/Minbyul/selfbiorag-7b-wo-medication_qa-sft) on an unknown dataset.
+This model is a fine-tuned version of [Minbyul/selfbiorag-7b-wo-medication_qa-sft](https://huggingface.co/Minbyul/selfbiorag-7b-wo-medication_qa-sft) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Logits/chosen: -0.3523
-- Logits/rejected: -0.3283
-- Logps/chosen: -679.6155
-- Logps/rejected: -1442.4718
-- Loss: 0.2756
+- Loss: 0.2759
+- Rewards/chosen: -1.2305
+- Rewards/rejected: -7.1000
 - Rewards/accuracies: 0.8920
-- Rewards/chosen: -1.2277
-- Rewards/margins: 5.8714
-- Rewards/rejected: -7.0991
+- Rewards/margins: 5.8695
+- Logps/rejected: -1442.5582
+- Logps/chosen: -679.8936
+- Logits/rejected: -0.3285
+- Logits/chosen: -0.3524
 
 ## Model description
 

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -0.3525831401348114,
-    "eval_logits/rejected": -0.32840025424957275,
-    "eval_logps/chosen": -679.7597045898438,
-    "eval_logps/rejected": -1442.6944580078125,
-    "eval_loss": 0.275810569524765,
-    "eval_rewards/accuracies": 0.8948863744735718,
-    "eval_rewards/chosen": -1.229145884513855,
-    "eval_rewards/margins": 5.8722147941589355,
-    "eval_rewards/rejected": -7.101361274719238,
-    "eval_runtime": 172.6817,
+    "eval_logits/chosen": -0.35237225890159607,
+    "eval_logits/rejected": -0.32851526141166687,
+    "eval_logps/chosen": -679.8935546875,
+    "eval_logps/rejected": -1442.5582275390625,
+    "eval_loss": 0.27591243386268616,
+    "eval_rewards/accuracies": 0.8920454382896423,
+    "eval_rewards/chosen": -1.230484962463379,
+    "eval_rewards/margins": 5.869514465332031,
+    "eval_rewards/rejected": -7.099999904632568,
+    "eval_runtime": 171.0958,
     "eval_samples": 1392,
-    "eval_samples_per_second": 8.061,
-    "eval_steps_per_second": 0.255,
+    "eval_samples_per_second": 8.136,
+    "eval_steps_per_second": 0.257,
     "train_loss": 0.0030385508506429234,
     "train_runtime": 144.6456,
     "train_samples": 19761,

config.json

@@ -23,6 +23,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32016
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -0.3525831401348114,
-    "eval_logits/rejected": -0.32840025424957275,
-    "eval_logps/chosen": -679.7597045898438,
-    "eval_logps/rejected": -1442.6944580078125,
-    "eval_loss": 0.275810569524765,
-    "eval_rewards/accuracies": 0.8948863744735718,
-    "eval_rewards/chosen": -1.229145884513855,
-    "eval_rewards/margins": 5.8722147941589355,
-    "eval_rewards/rejected": -7.101361274719238,
-    "eval_runtime": 172.6817,
+    "eval_logits/chosen": -0.35237225890159607,
+    "eval_logits/rejected": -0.32851526141166687,
+    "eval_logps/chosen": -679.8935546875,
+    "eval_logps/rejected": -1442.5582275390625,
+    "eval_loss": 0.27591243386268616,
+    "eval_rewards/accuracies": 0.8920454382896423,
+    "eval_rewards/chosen": -1.230484962463379,
+    "eval_rewards/margins": 5.869514465332031,
+    "eval_rewards/rejected": -7.099999904632568,
+    "eval_runtime": 171.0958,
     "eval_samples": 1392,
-    "eval_samples_per_second": 8.061,
-    "eval_steps_per_second": 0.255
+    "eval_samples_per_second": 8.136,
+    "eval_steps_per_second": 0.257
 }