Model save

README.md

@@ -27,7 +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/ziyaow-university-of-maryland/Qwen2.5-FL-GRPO/runs/h3yagvyr) 
+[<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/ziyaow-university-of-maryland/Qwen2.5-FL-GRPO/runs/gb50qdmv) 
 
 
 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

@@ -1,9 +1,9 @@
 {
     "client_id": 0,
     "total_flos": 0.0,
-    "train_loss": 0.013306360070904097,
-    "train_runtime": 224.5809,
+    "train_loss": 0.00680889127155145,
+    "train_runtime": 227.697,
     "train_samples": 100,
-    "train_samples_per_second": 0.445,
+    "train_samples_per_second": 0.439,
     "train_steps_per_second": 0.018
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69422b2b44b3dcbd45e8d0a735da7cc6d0570de2efd7930f1714d6c98362a0e9
+oid sha256:e3648f0a5833e6cd44b4ec131d680ba1c177f67845f190c4b370e7f7ce1aa243
 size 3087467144

train_client_0_results.json

@@ -1,9 +1,9 @@
 {
     "client_id": 0,
     "total_flos": 0.0,
-    "train_loss": 0.013306360070904097,
-    "train_runtime": 224.5809,
+    "train_loss": 0.00680889127155145,
+    "train_runtime": 227.697,
     "train_samples": 100,
-    "train_samples_per_second": 0.445,
+    "train_samples_per_second": 0.439,
     "train_steps_per_second": 0.018
 }

trainer_state.json

@@ -24,7 +24,7 @@
       "completions/min_terminated_length": 44.0,
       "epoch": 0.3076923076923077,
       "frac_reward_zero_std": 0.28125,
-      "grad_norm": 0.2417769552255137,
+      "grad_norm": 0.24184319600046375,
       "kl": 0.00023603439331054688,
       "learning_rate": 0.0,
       "loss": 0.0155,
@@ -45,36 +45,36 @@
       "clip_ratio/low_mean": 0.0,
       "clip_ratio/low_min": 0.0,
       "clip_ratio/region_mean": 0.0,
-      "completions/clipped_ratio": -6.421875,
+      "completions/clipped_ratio": -6.453125,
       "completions/max_length": 1024.0,
-      "completions/max_terminated_length": 1022.5,
-      "completions/mean_length": 641.396484375,
-      "completions/mean_terminated_length": 611.4868469238281,
-      "completions/min_length": 82.5,
-      "completions/min_terminated_length": 82.5,
+      "completions/max_terminated_length": 1020.0,
+      "completions/mean_length": 646.392578125,
+      "completions/mean_terminated_length": 618.6748657226562,
+      "completions/min_length": 71.5,
+      "completions/min_terminated_length": 71.5,
       "epoch": 0.9230769230769231,
       "frac_reward_zero_std": 0.171875,
-      "kl": 0.006296515464782715,
-      "num_tokens": 1230395.0,
-      "reward": 0.32421875,
-      "reward_std": 0.21514922380447388,
-      "rewards/accuracy_reward/mean": 0.0224609375,
-      "rewards/accuracy_reward/std": 0.14353204891085625,
-      "rewards/format_reward/mean": 0.0029296875,
-      "rewards/format_reward/std": 0.038198307156562805,
-      "rewards/tag_count_reward/mean": 0.298828125,
-      "rewards/tag_count_reward/std": 0.1979760304093361,
+      "kl": 0.006029486656188965,
+      "num_tokens": 1235511.0,
+      "reward": 0.33740234375,
+      "reward_std": 0.22876286506652832,
+      "rewards/accuracy_reward/mean": 0.0263671875,
+      "rewards/accuracy_reward/std": 0.1589299514889717,
+      "rewards/format_reward/mean": 0.009765625,
+      "rewards/format_reward/std": 0.09631745889782906,
+      "rewards/tag_count_reward/mean": 0.30126953125,
+      "rewards/tag_count_reward/std": 0.19553955644369125,
       "step": 3,
       "total_flos": 0.0,
-      "train_loss": 0.013306360070904097,
-      "train_runtime": 224.5809,
-      "train_samples_per_second": 0.445,
+      "train_loss": 0.00680889127155145,
+      "train_runtime": 227.697,
+      "train_samples_per_second": 0.439,
       "train_steps_per_second": 0.018
     }
   ],
   "logging_steps": 10,
   "max_steps": 4,
-  "num_input_tokens_seen": 1230395,
+  "num_input_tokens_seen": 1235511,
   "num_train_epochs": 1,
   "save_steps": 500,
   "stateful_callbacks": {

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:23052ae58d9196c53e2cbe990c47439852faef94315ff1e62b2164bb6fc5c7ab
+oid sha256:db1a45478861409988aee4fb61eed49d7249e737837ff0471082a898297f8d9c
 size 8440