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 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: 0.4963
-- Rewards/chosen: -1.3653
-- Rewards/rejected: -2.3944
-- Rewards/accuracies: 0.7639
-- Rewards/margins: 1.0292
-- Logps/rejected: -499.6516
-- Logps/chosen: -418.4965
-- Logits/rejected: 1.1202
-- Logits/chosen: 0.2514
+- Rewards/chosen: -1.3619
+- Rewards/rejected: -2.3893
+- Rewards/accuracies: 0.7619
+- Rewards/margins: 1.0274
+- Logps/rejected: -499.1414
+- Logps/chosen: -418.1593
+- Logits/rejected: 1.1001
+- Logits/chosen: 0.2372
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 0.9994767137624281,
+    "eval_logits/chosen": 0.23715664446353912,
+    "eval_logits/rejected": 1.1001231670379639,
+    "eval_logps/chosen": -418.1593017578125,
+    "eval_logps/rejected": -499.1413879394531,
+    "eval_loss": 0.4963093400001526,
+    "eval_rewards/accuracies": 0.761904776096344,
+    "eval_rewards/chosen": -1.3619133234024048,
+    "eval_rewards/margins": 1.0274325609207153,
+    "eval_rewards/rejected": -2.38934588432312,
+    "eval_runtime": 185.5011,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 10.782,
+    "eval_steps_per_second": 0.34,
     "total_flos": 0.0,
     "train_loss": 0.1815465917137905,
     "train_runtime": 7518.522,

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.41.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 0.9994767137624281,
+    "eval_logits/chosen": 0.23715664446353912,
+    "eval_logits/rejected": 1.1001231670379639,
+    "eval_logps/chosen": -418.1593017578125,
+    "eval_logps/rejected": -499.1413879394531,
+    "eval_loss": 0.4963093400001526,
+    "eval_rewards/accuracies": 0.761904776096344,
+    "eval_rewards/chosen": -1.3619133234024048,
+    "eval_rewards/margins": 1.0274325609207153,
+    "eval_rewards/rejected": -2.38934588432312,
+    "eval_runtime": 185.5011,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 10.782,
+    "eval_steps_per_second": 0.34
+}

runs/Jun21_05-38-58_ishant-gpu/events.out.tfevents.1718957941.ishant-gpu.19428.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72f31c47b46436e9b602c1225581538dd58e985742a7dcb8665143e0ce968c24
+size 828