End of training

README.md

@@ -2,10 +2,16 @@
 license: apache-2.0
 base_model: alignment-handbook/zephyr-7b-sft-full
 tags:
+- alignment-handbook
+- trl
+- dpo
+- generated_from_trainer
 - trl
 - dpo
 - alignment-handbook
 - generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: zephyr-7b-dpo-full-ultrabin-high-bleu
   results: []
@@ -16,17 +22,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-dpo-full-ultrabin-high-bleu
 
-This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown 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: 0.6311
-- Rewards/chosen: -0.3272
-- Rewards/rejected: -0.5547
-- Rewards/accuracies: 0.6875
-- Rewards/margins: 0.2276
-- Logps/rejected: -318.1367
-- Logps/chosen: -295.3459
-- Logits/rejected: -2.3484
-- Logits/chosen: -2.3932
+- Loss: 0.6250
+- Rewards/chosen: -0.3165
+- Rewards/rejected: -0.5606
+- Rewards/accuracies: 0.6953
+- Rewards/margins: 0.2441
+- Logps/rejected: -318.7193
+- Logps/chosen: -294.2789
+- Logits/rejected: -2.3036
+- Logits/chosen: -2.3509
 
 ## Model description
 

all_results.json

@@ -1,18 +1,18 @@
 {
     "epoch": 0.9965156794425087,
-    "eval_logits/chosen": -2.281348943710327,
-    "eval_logits/rejected": -2.2318127155303955,
-    "eval_logps/chosen": -292.6180114746094,
-    "eval_logps/rejected": -316.9007568359375,
-    "eval_loss": 0.6281932592391968,
-    "eval_rewards/accuracies": 0.703125,
-    "eval_rewards/chosen": -0.2998806834220886,
-    "eval_rewards/margins": 0.2425060272216797,
-    "eval_rewards/rejected": -0.5423867106437683,
-    "eval_runtime": 104.9019,
+    "eval_logits/chosen": -2.3508927822113037,
+    "eval_logits/rejected": -2.3035717010498047,
+    "eval_logps/chosen": -294.2789306640625,
+    "eval_logps/rejected": -318.7192687988281,
+    "eval_loss": 0.624984860420227,
+    "eval_rewards/accuracies": 0.6953125,
+    "eval_rewards/chosen": -0.316490113735199,
+    "eval_rewards/margins": 0.24408146739006042,
+    "eval_rewards/rejected": -0.5605716109275818,
+    "eval_runtime": 105.1694,
     "eval_samples": 2000,
-    "eval_samples_per_second": 19.065,
-    "eval_steps_per_second": 0.305,
+    "eval_samples_per_second": 19.017,
+    "eval_steps_per_second": 0.304,
     "total_flos": 0.0,
     "train_loss": 0.6505647914392965,
     "train_runtime": 3852.398,

config.json

@@ -22,6 +22,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 0.9965156794425087,
-    "eval_logits/chosen": -2.281348943710327,
-    "eval_logits/rejected": -2.2318127155303955,
-    "eval_logps/chosen": -292.6180114746094,
-    "eval_logps/rejected": -316.9007568359375,
-    "eval_loss": 0.6281932592391968,
-    "eval_rewards/accuracies": 0.703125,
-    "eval_rewards/chosen": -0.2998806834220886,
-    "eval_rewards/margins": 0.2425060272216797,
-    "eval_rewards/rejected": -0.5423867106437683,
-    "eval_runtime": 104.9019,
+    "eval_logits/chosen": -2.3508927822113037,
+    "eval_logits/rejected": -2.3035717010498047,
+    "eval_logps/chosen": -294.2789306640625,
+    "eval_logps/rejected": -318.7192687988281,
+    "eval_loss": 0.624984860420227,
+    "eval_rewards/accuracies": 0.6953125,
+    "eval_rewards/chosen": -0.316490113735199,
+    "eval_rewards/margins": 0.24408146739006042,
+    "eval_rewards/rejected": -0.5605716109275818,
+    "eval_runtime": 105.1694,
     "eval_samples": 2000,
-    "eval_samples_per_second": 19.065,
-    "eval_steps_per_second": 0.305
+    "eval_samples_per_second": 19.017,
+    "eval_steps_per_second": 0.304
 }