Scientific Reasoning on SciKnowEval Physics
63.39Mean@16GRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| GRPOModel Size=Qwen3-8B, Training Domain=Material2026.05 | 63.39 | |
| GRPOModel Size=Qwen3-8B, Training Domain=Chemistry2026.05 | 61.8 | |
| GRPOModel Size=Qwen3-8B, Training Domain=Biology2026.05 | 60.89 | |
| GRPOModel Size=Qwen3-4B, Training Domain=Material2026.05 | 60.78 | |
| GRPOModel Size=Qwen3-8B, Training Domain=ToolUse2026.05 | 57.76 | |
| GRPOModel Size=Qwen3-4B, Training Domain=ToolUse2026.05 | 57.29 | |
| ROSDModel Size=Qwen3-8B, Training Domain=Biology2026.05 | 56.98 | |
| GRPOModel Size=Qwen3-4B, Training Domain=Biology2026.05 | 56.35 | |
| ROSDModel Size=Qwen3-8B, Training Domain=Material2026.05 | 55.57 | |
| ROSDModel Size=Qwen3-8B, Training Domain=ToolUse2026.05 | 54.3 | |
| GRPOModel Size=Qwen3-4B, Training Domain=Chemistry2026.05 | 49.71 | |
| ROSDModel Size=Qwen3-4B, Training Domain=Biology2026.05 | 49.58 | |
| ROSDModel Size=Qwen3-8B, Training Domain=Chemistry2026.05 | 49.09 | |
| ROSDModel Size=Qwen3-4B, Training Domain=ToolUse2026.05 | 46.25 | |
| SDPOModel Size=Qwen3-4B, Training Domain=Biology2026.05 | 45.68 | |
| ROSDModel Size=Qwen3-4B, Training Domain=Material2026.05 | 45.05 | |
| SDPOModel Size=Qwen3-8B, Training Domain=Material2026.05 | 42.76 | |
| SDPOModel Size=Qwen3-8B, Training Domain=Biology2026.05 | 40.36 | |
| SDPOModel Size=Qwen3-8B, Training Domain=ToolUse2026.05 | 40.34 | |
| ROSDModel Size=Qwen3-4B, Training Domain=Chemistry2026.05 | 39.95 | |
| SDPOModel Size=Qwen3-4B, Training Domain=Material2026.05 | 30.29 | |
| SDPOModel Size=Qwen3-8B, Training Domain=Chemistry2026.05 | 29.92 | |
| SDPOModel Size=Qwen3-4B, Training Domain=Chemistry2026.05 | 23.05 | |
| SDPOModel Size=Qwen3-4B, Training Domain=ToolUse2026.05 | 0.49 |