End of training

README.md

@@ -1,9 +1,15 @@
 ---
 base_model: Minbyul/selfbiorag-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: selfbiorag-7b-dpo-full-sft-wo-healthsearch_qa
   results: []
@@ -14,7 +20,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # selfbiorag-7b-dpo-full-sft-wo-healthsearch_qa
 
-This model is a fine-tuned version of [Minbyul/selfbiorag-7b-wo-healthsearch_qa-sft](https://huggingface.co/Minbyul/selfbiorag-7b-wo-healthsearch_qa-sft) on an unknown dataset.
+This model is a fine-tuned version of [Minbyul/selfbiorag-7b-wo-healthsearch_qa-sft](https://huggingface.co/Minbyul/selfbiorag-7b-wo-healthsearch_qa-sft) on the HuggingFaceH4/ultrafeedback_binarized dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.4465
+- Rewards/chosen: -0.5929
+- Rewards/rejected: -1.6772
+- Rewards/accuracies: 0.7846
+- Rewards/margins: 1.0843
+- Logps/rejected: -1480.8429
+- Logps/chosen: -984.8102
+- Logits/rejected: -3.4642
+- Logits/chosen: -2.6475
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.6475114822387695,
+    "eval_logits/rejected": -3.464193344116211,
+    "eval_logps/chosen": -984.8101806640625,
+    "eval_logps/rejected": -1480.8428955078125,
+    "eval_loss": 0.4465055763721466,
+    "eval_rewards/accuracies": 0.7845849990844727,
+    "eval_rewards/chosen": -0.5929392576217651,
+    "eval_rewards/margins": 1.0842887163162231,
+    "eval_rewards/rejected": -1.6772277355194092,
+    "eval_runtime": 1610.6329,
+    "eval_samples": 16161,
+    "eval_samples_per_second": 10.034,
+    "eval_steps_per_second": 0.314,
     "train_loss": 0.2915988182410216,
     "train_runtime": 1026.489,
     "train_samples": 4992,

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

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -2.6475114822387695,
+    "eval_logits/rejected": -3.464193344116211,
+    "eval_logps/chosen": -984.8101806640625,
+    "eval_logps/rejected": -1480.8428955078125,
+    "eval_loss": 0.4465055763721466,
+    "eval_rewards/accuracies": 0.7845849990844727,
+    "eval_rewards/chosen": -0.5929392576217651,
+    "eval_rewards/margins": 1.0842887163162231,
+    "eval_rewards/rejected": -1.6772277355194092,
+    "eval_runtime": 1610.6329,
+    "eval_samples": 16161,
+    "eval_samples_per_second": 10.034,
+    "eval_steps_per_second": 0.314
+}