Model save

README.md

@@ -1,11 +1,9 @@
 ---
 base_model: Qwen/Qwen2.5-3B-Instruct
-datasets: open-r1/verifiable-coding-problems-python
 library_name: transformers
 model_name: Qwen2.5-3B-Open-R1-Code-GRPO
 tags:
 - generated_from_trainer
-- open-r1
 - trl
 - grpo
 licence: license
@@ -13,7 +11,7 @@ licence: license
 
 # Model Card for Qwen2.5-3B-Open-R1-Code-GRPO
 
-This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct) on the [open-r1/verifiable-coding-problems-python](https://huggingface.co/datasets/open-r1/verifiable-coding-problems-python) dataset.
+This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start
@@ -29,7 +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/chenyukang2020-nvidia/huggingface/runs/4kiw3xot) 
+[<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/chenyukang2020-nvidia/huggingface/runs/87jsozpo) 
 
 
 This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 1.3396084333325358e-07,
-    "train_runtime": 69.6904,
+    "train_loss": 7.054507823697492e-06,
+    "train_runtime": 74.8551,
     "train_samples": 35735,
-    "train_samples_per_second": 3673.392,
-    "train_steps_per_second": 7.175
+    "train_samples_per_second": 3419.94,
+    "train_steps_per_second": 6.68
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 1.3396084333325358e-07,
-    "train_runtime": 69.6904,
+    "train_loss": 7.054507823697492e-06,
+    "train_runtime": 74.8551,
     "train_samples": 35735,
-    "train_samples_per_second": 3673.392,
-    "train_steps_per_second": 7.175
+    "train_samples_per_second": 3419.94,
+    "train_steps_per_second": 6.68
 }

trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.4557868815760018,
+  "epoch": 0.45668233713901946,
   "eval_steps": 500,
-  "global_step": 509,
+  "global_step": 510,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -14262,18 +14262,46 @@
       "step": 509
     },
     {
-      "epoch": 0.4557868815760018,
-      "step": 509,
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completions/clipped_ratio": -7.0,
+      "completions/max_length": 1611.0,
+      "completions/max_terminated_length": 1611.0,
+      "completions/mean_length": 572.6640625,
+      "completions/mean_terminated_length": 572.6640625,
+      "completions/min_length": 170.0,
+      "completions/min_terminated_length": 170.0,
+      "epoch": 0.45668233713901946,
+      "frac_reward_zero_std": 0.96875,
+      "grad_norm": 0.05233213840958361,
+      "kl": 0.00714874267578125,
+      "learning_rate": 5.038223531225739e-07,
+      "loss": 0.0036,
+      "num_tokens": 305889150.0,
+      "reward": 0.09980468451976776,
+      "reward_std": 0.0007812500116415322,
+      "rewards/code_reward/mean": 0.0,
+      "rewards/code_reward/std": 0.0,
+      "rewards/format_reward/mean": 0.998046875,
+      "rewards/format_reward/std": 0.04419417306780815,
+      "step": 510
+    },
+    {
+      "epoch": 0.45668233713901946,
+      "step": 510,
       "total_flos": 0.0,
-      "train_loss": 1.3396084333325358e-07,
-      "train_runtime": 69.6904,
-      "train_samples_per_second": 3673.392,
-      "train_steps_per_second": 7.175
+      "train_loss": 7.054507823697492e-06,
+      "train_runtime": 74.8551,
+      "train_samples_per_second": 3419.94,
+      "train_steps_per_second": 6.68
     }
   ],
   "logging_steps": 1,
   "max_steps": 500,
-  "num_input_tokens_seen": 305322634,
+  "num_input_tokens_seen": 305889150,
   "num_train_epochs": 1,
   "save_steps": 50,
   "stateful_callbacks": {