Model save

README.md

@@ -1,6 +1,5 @@
 ---
 base_model: JayHyeon/Qwen2.5-0.5B-SFT-2e-5-2ep
-datasets: trl-lib/ultrafeedback_binarized
 library_name: transformers
 model_name: Qwen2.5-0.5B-SFT-2e-5-2ep-MDPO_5e-7-3ep_0alp_0lam
 tags:
@@ -12,7 +11,7 @@ licence: license
 
 # Model Card for Qwen2.5-0.5B-SFT-2e-5-2ep-MDPO_5e-7-3ep_0alp_0lam
 
-This model is a fine-tuned version of [JayHyeon/Qwen2.5-0.5B-SFT-2e-5-2ep](https://huggingface.co/JayHyeon/Qwen2.5-0.5B-SFT-2e-5-2ep) on the [trl-lib/ultrafeedback_binarized](https://huggingface.co/datasets/trl-lib/ultrafeedback_binarized) dataset.
+This model is a fine-tuned version of [JayHyeon/Qwen2.5-0.5B-SFT-2e-5-2ep](https://huggingface.co/JayHyeon/Qwen2.5-0.5B-SFT-2e-5-2ep).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start
@@ -28,7 +27,7 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/bonin147/huggingface/runs/bkjep6bm)
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/bonin147/huggingface/runs/0y2rv565)
 
 This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
 

all_results.json

@@ -1,22 +1,22 @@
 {
     "epoch": 2.999098751126561,
-    "eval_kl/chosen": 0.00026274315314367414,
-    "eval_kl/rejected": 0.00014435610501095653,
+    "eval_kl/chosen": 0.000334249489242211,
+    "eval_kl/rejected": 0.00022987440752331167,
     "eval_logits/chosen": -1.4296875,
-    "eval_logits/rejected": -1.390625,
+    "eval_logits/rejected": -1.3984375,
     "eval_logps/chosen": -322.0,
     "eval_logps/rejected": -298.0,
-    "eval_loss": 0.6893105506896973,
+    "eval_loss": 0.688232421875,
     "eval_nll_loss": 1.046875,
     "eval_ref_probs/chosen": 0.0038271306548267603,
     "eval_ref_probs/rejected": 0.0034569373819977045,
-    "eval_rejected_term/max": 0.00014435610501095653,
-    "eval_rejected_term/min": 0.00014435610501095653,
-    "eval_rewards/accuracies": 0.6520000100135803,
-    "eval_rewards/chosen": -0.00738525390625,
-    "eval_rewards/margins": 0.134765625,
-    "eval_rewards/rejected": -0.1416015625,
-    "eval_runtime": 25.3586,
-    "eval_samples_per_second": 39.434,
-    "eval_steps_per_second": 9.859
+    "eval_rejected_term/max": 0.00022987440752331167,
+    "eval_rejected_term/min": 0.00022987440752331167,
+    "eval_rewards/accuracies": 0.6079999804496765,
+    "eval_rewards/chosen": 0.01031494140625,
+    "eval_rewards/margins": 0.1435546875,
+    "eval_rewards/rejected": -0.1337890625,
+    "eval_runtime": 25.3139,
+    "eval_samples_per_second": 39.504,
+    "eval_steps_per_second": 9.876
 }

eval_results.json

@@ -1,22 +1,22 @@
 {
     "epoch": 2.999098751126561,
-    "eval_kl/chosen": 0.00026274315314367414,
-    "eval_kl/rejected": 0.00014435610501095653,
+    "eval_kl/chosen": 0.000334249489242211,
+    "eval_kl/rejected": 0.00022987440752331167,
     "eval_logits/chosen": -1.4296875,
-    "eval_logits/rejected": -1.390625,
+    "eval_logits/rejected": -1.3984375,
     "eval_logps/chosen": -322.0,
     "eval_logps/rejected": -298.0,
-    "eval_loss": 0.6893105506896973,
+    "eval_loss": 0.688232421875,
     "eval_nll_loss": 1.046875,
     "eval_ref_probs/chosen": 0.0038271306548267603,
     "eval_ref_probs/rejected": 0.0034569373819977045,
-    "eval_rejected_term/max": 0.00014435610501095653,
-    "eval_rejected_term/min": 0.00014435610501095653,
-    "eval_rewards/accuracies": 0.6520000100135803,
-    "eval_rewards/chosen": -0.00738525390625,
-    "eval_rewards/margins": 0.134765625,
-    "eval_rewards/rejected": -0.1416015625,
-    "eval_runtime": 25.3586,
-    "eval_samples_per_second": 39.434,
-    "eval_steps_per_second": 9.859
+    "eval_rejected_term/max": 0.00022987440752331167,
+    "eval_rejected_term/min": 0.00022987440752331167,
+    "eval_rewards/accuracies": 0.6079999804496765,
+    "eval_rewards/chosen": 0.01031494140625,
+    "eval_rewards/margins": 0.1435546875,
+    "eval_rewards/rejected": -0.1337890625,
+    "eval_runtime": 25.3139,
+    "eval_samples_per_second": 39.504,
+    "eval_steps_per_second": 9.876
 }