End of training

README.md

@@ -2,9 +2,15 @@
 license: apache-2.0
 base_model: mistralai/Mistral-7B-v0.1
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: mistral-7b-dpo-full-wo-live_qa-ep3
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # mistral-7b-dpo-full-wo-live_qa-ep3
 
-This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
+This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3346
-- Rewards/chosen: -1.2129
-- Rewards/rejected: -5.3286
-- Rewards/accuracies: 0.7917
-- Rewards/margins: 4.1158
-- Logps/rejected: -1444.5776
-- Logps/chosen: -972.7769
-- Logits/rejected: -3.3074
-- Logits/chosen: -3.3846
+- Loss: 0.2304
+- Rewards/chosen: -1.4964
+- Rewards/rejected: -8.5718
+- Rewards/accuracies: 0.8472
+- Rewards/margins: 7.0755
+- Logps/rejected: -1768.8982
+- Logps/chosen: -1001.1289
+- Logits/rejected: -3.1372
+- Logits/chosen: -3.3307
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -3.3306896686553955,
+    "eval_logits/rejected": -3.137249708175659,
+    "eval_logps/chosen": -1001.12890625,
+    "eval_logps/rejected": -1768.898193359375,
+    "eval_loss": 0.2304462045431137,
+    "eval_rewards/accuracies": 0.8472222089767456,
+    "eval_rewards/chosen": -1.4963774681091309,
+    "eval_rewards/margins": 7.075459003448486,
+    "eval_rewards/rejected": -8.5718355178833,
+    "eval_runtime": 34.9747,
+    "eval_samples": 284,
+    "eval_samples_per_second": 8.12,
+    "eval_steps_per_second": 0.257,
     "train_loss": 0.2672084366101675,
     "train_runtime": 3083.1249,
     "train_samples": 12328,

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": -3.3306896686553955,
+    "eval_logits/rejected": -3.137249708175659,
+    "eval_logps/chosen": -1001.12890625,
+    "eval_logps/rejected": -1768.898193359375,
+    "eval_loss": 0.2304462045431137,
+    "eval_rewards/accuracies": 0.8472222089767456,
+    "eval_rewards/chosen": -1.4963774681091309,
+    "eval_rewards/margins": 7.075459003448486,
+    "eval_rewards/rejected": -8.5718355178833,
+    "eval_runtime": 34.9747,
+    "eval_samples": 284,
+    "eval_samples_per_second": 8.12,
+    "eval_steps_per_second": 0.257
+}