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/87u2z6ka) 
+[<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/2ky2sf2l) 
 
 
 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": 0.011905932798981667,
-    "train_runtime": 227.3014,
+    "train_loss": 0.015650476639469463,
+    "train_runtime": 225.343,
     "train_samples": 100,
-    "train_samples_per_second": 0.44,
+    "train_samples_per_second": 0.444,
     "train_steps_per_second": 0.018
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dd924a11b4c220f385b51ffa522daea7c9f3d850e31b162bb5661df483c6d3ee
+oid sha256:0bf4e2a528040938909343db580320818effc31b2425fb1bcfb21287eadc6f74
 size 3087467144

train_client_0_results.json

@@ -1,9 +1,9 @@
 {
     "client_id": 0,
     "total_flos": 0.0,
-    "train_loss": 0.011905932798981667,
-    "train_runtime": 227.3014,
+    "train_loss": 0.015650476639469463,
+    "train_runtime": 225.343,
     "train_samples": 100,
-    "train_samples_per_second": 0.44,
+    "train_samples_per_second": 0.444,
     "train_steps_per_second": 0.018
 }

trainer_state.json

@@ -24,6 +24,7 @@
       "completions/min_terminated_length": 44.0,
       "epoch": 0.3076923076923077,
       "frac_reward_zero_std": 0.28125,
+      "grad_norm": 0.24192178551600718,
       "kl": 0.00023603439331054688,
       "learning_rate": 0.0,
       "loss": 0.0155,
@@ -44,36 +45,36 @@
       "clip_ratio/low_mean": 0.0,
       "clip_ratio/low_min": 0.0,
       "clip_ratio/region_mean": 0.0,
-      "completions/clipped_ratio": -6.3515625,
+      "completions/clipped_ratio": -6.359375,
       "completions/max_length": 1024.0,
-      "completions/max_terminated_length": 1021.5,
-      "completions/mean_length": 632.998046875,
-      "completions/mean_terminated_length": 598.0208129882812,
-      "completions/min_length": 62.5,
-      "completions/min_terminated_length": 62.5,
+      "completions/max_terminated_length": 1022.5,
+      "completions/mean_length": 645.326171875,
+      "completions/mean_terminated_length": 612.1986083984375,
+      "completions/min_length": 71.5,
+      "completions/min_terminated_length": 71.5,
       "epoch": 0.9230769230769231,
-      "frac_reward_zero_std": 0.421875,
-      "kl": 0.00025832653045654297,
-      "num_tokens": 1221795.0,
-      "reward": 0.075439453125,
-      "reward_std": 0.17266184091567993,
-      "rewards/accuracy_reward/mean": 0.033203125,
-      "rewards/accuracy_reward/std": 0.17934183776378632,
-      "rewards/format_reward/mean": 0.0068359375,
-      "rewards/format_reward/std": 0.0822625607252121,
-      "rewards/tag_count_reward/mean": 0.035400390625,
-      "rewards/tag_count_reward/std": 0.15237901359796524,
+      "frac_reward_zero_std": 0.171875,
+      "kl": 0.006090521812438965,
+      "num_tokens": 1234419.0,
+      "reward": 0.336181640625,
+      "reward_std": 0.23990178108215332,
+      "rewards/accuracy_reward/mean": 0.025390625,
+      "rewards/accuracy_reward/std": 0.15544121712446213,
+      "rewards/format_reward/mean": 0.009765625,
+      "rewards/format_reward/std": 0.09631745889782906,
+      "rewards/tag_count_reward/mean": 0.301025390625,
+      "rewards/tag_count_reward/std": 0.2050374299287796,
       "step": 3,
       "total_flos": 0.0,
-      "train_loss": 0.011905932798981667,
-      "train_runtime": 227.3014,
-      "train_samples_per_second": 0.44,
+      "train_loss": 0.015650476639469463,
+      "train_runtime": 225.343,
+      "train_samples_per_second": 0.444,
       "train_steps_per_second": 0.018
     }
   ],
   "logging_steps": 10,
   "max_steps": 4,
-  "num_input_tokens_seen": 1221795,
+  "num_input_tokens_seen": 1234419,
   "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:b4363a974af3787ab17fba919ba477e62fe785073809f977f99e63f3267aa284
+oid sha256:8a1dfd6de7736d0ffce3223e2f3397ac8e4181cc0bd5efa5aa42d7644b85fad9
 size 8440