End of training

README.md

@@ -2,9 +2,15 @@
 license: llama2
 base_model: epfl-llm/meditron-7b
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: meditron-7b-dpo-full-wo-medication_qa-ep3
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # meditron-7b-dpo-full-wo-medication_qa-ep3
 
-This model is a fine-tuned version of [epfl-llm/meditron-7b](https://huggingface.co/epfl-llm/meditron-7b) on an unknown dataset.
+This model is a fine-tuned version of [epfl-llm/meditron-7b](https://huggingface.co/epfl-llm/meditron-7b) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6293
-- Rewards/chosen: -0.0145
-- Rewards/rejected: -0.1627
-- Rewards/accuracies: 0.7797
-- Rewards/margins: 0.1482
-- Logps/rejected: -1615.9989
-- Logps/chosen: -955.1961
-- Logits/rejected: -0.8722
-- Logits/chosen: -0.7069
+- Loss: 0.5850
+- Rewards/chosen: -0.2075
+- Rewards/rejected: -0.5850
+- Rewards/accuracies: 0.7881
+- Rewards/margins: 0.3775
+- Logps/rejected: -1658.2283
+- Logps/chosen: -974.4987
+- Logits/rejected: -0.8799
+- Logits/chosen: -0.7592
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -0.7592486143112183,
+    "eval_logits/rejected": -0.87989342212677,
+    "eval_logps/chosen": -974.4986572265625,
+    "eval_logps/rejected": -1658.228271484375,
+    "eval_loss": 0.5849764347076416,
+    "eval_rewards/accuracies": 0.7881355881690979,
+    "eval_rewards/chosen": -0.20753850042819977,
+    "eval_rewards/margins": 0.3774545192718506,
+    "eval_rewards/rejected": -0.5849930644035339,
+    "eval_runtime": 215.1803,
+    "eval_samples": 1864,
+    "eval_samples_per_second": 8.663,
+    "eval_steps_per_second": 0.274,
     "train_loss": 0.5914445311983646,
     "train_runtime": 2922.1625,
     "train_samples": 11603,

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": 32017
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -0.7592486143112183,
+    "eval_logits/rejected": -0.87989342212677,
+    "eval_logps/chosen": -974.4986572265625,
+    "eval_logps/rejected": -1658.228271484375,
+    "eval_loss": 0.5849764347076416,
+    "eval_rewards/accuracies": 0.7881355881690979,
+    "eval_rewards/chosen": -0.20753850042819977,
+    "eval_rewards/margins": 0.3774545192718506,
+    "eval_rewards/rejected": -0.5849930644035339,
+    "eval_runtime": 215.1803,
+    "eval_samples": 1864,
+    "eval_samples_per_second": 8.663,
+    "eval_steps_per_second": 0.274
+}