End of training

README.md

@@ -2,9 +2,15 @@
 license: apache-2.0
 base_model: alignment-handbook/zephyr-7b-sft-full
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: zephyr-7b-dpo-full
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-dpo-full
 
-This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
+This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4917
-- Rewards/chosen: 21.2230
-- Rewards/rejected: 6.3692
+- Loss: 0.4929
+- Rewards/chosen: 21.1860
+- Rewards/rejected: 6.2518
 - Rewards/accuracies: 0.7344
-- Rewards/margins: 14.8539
-- Logps/rejected: -256.2980
-- Logps/chosen: -241.3705
-- Logits/rejected: -2.7037
-- Logits/chosen: -2.7315
+- Rewards/margins: 14.9342
+- Logps/rejected: -256.4154
+- Logps/chosen: -241.4075
+- Logits/rejected: -2.7091
+- Logits/chosen: -2.7366
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.7366418838500977,
+    "eval_logits/rejected": -2.709143877029419,
+    "eval_logps/chosen": -241.407470703125,
+    "eval_logps/rejected": -256.4153747558594,
+    "eval_loss": 0.4928562045097351,
+    "eval_rewards/accuracies": 0.734375,
+    "eval_rewards/chosen": 21.185983657836914,
+    "eval_rewards/margins": 14.934151649475098,
+    "eval_rewards/rejected": 6.251833915710449,
+    "eval_runtime": 96.1252,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 20.806,
+    "eval_steps_per_second": 0.333,
     "train_loss": 0.5147256711536871,
     "train_runtime": 7551.132,
     "train_samples": 61134,

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.38.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -2.689567804336548,
-    "eval_logits/rejected": -2.664588451385498,
-    "eval_logps/chosen": -230.68809509277344,
-    "eval_logps/rejected": -235.1542205810547,
-    "eval_loss": 0.27835845947265625,
-    "eval_rewards/accuracies": 0.57421875,
-    "eval_rewards/chosen": 31.905372619628906,
-    "eval_rewards/margins": 4.392405033111572,
-    "eval_rewards/rejected": 27.51296615600586,
-    "eval_runtime": 96.4814,
+    "eval_logits/chosen": -2.7366418838500977,
+    "eval_logits/rejected": -2.709143877029419,
+    "eval_logps/chosen": -241.407470703125,
+    "eval_logps/rejected": -256.4153747558594,
+    "eval_loss": 0.4928562045097351,
+    "eval_rewards/accuracies": 0.734375,
+    "eval_rewards/chosen": 21.185983657836914,
+    "eval_rewards/margins": 14.934151649475098,
+    "eval_rewards/rejected": 6.251833915710449,
+    "eval_runtime": 96.1252,
     "eval_samples": 2000,
-    "eval_samples_per_second": 20.729,
-    "eval_steps_per_second": 0.332
+    "eval_samples_per_second": 20.806,
+    "eval_steps_per_second": 0.333
 }