Scientific Reasoning on SciKnowEval Chemistry
94.3mean@16DRIFT
Evaluation Results
| Method | Links | |
|---|---|---|
| DRIFTModel=Qwen3-8B2026.06 | 94.3 | |
| GRPOModel=Qwen3-8B2026.06 | 93.5 | |
| SDPOModel=Qwen3-8B2026.06 | 89.4 | |
| GRPOModel Size=Qwen3-8B, Training Domain=Material2026.05 | 46.87 | |
| GRPOModel Size=Qwen3-8B, Training Domain=Physics2026.05 | 46.12 | |
| ROSDModel Size=Qwen3-4B, Training Domain=Material2026.05 | 45.48 | |
| ROSDModel Size=Qwen3-4B, Training Domain=ToolUse2026.05 | 45.43 | |
| GRPOModel Size=Qwen3-4B, Training Domain=Biology2026.05 | 45.23 | |
| SDPOModel Size=Qwen3-8B, Training Domain=ToolUse2026.05 | 45.18 | |
| ROSDModel Size=Qwen3-8B, Training Domain=ToolUse2026.05 | 45.03 | |
| SDPOModel Size=Qwen3-8B, Training Domain=Material2026.05 | 44.07 | |
| SDPOModel Size=Qwen3-8B, Training Domain=Physics2026.05 | 44 | |
| GRPOModel Size=Qwen3-4B, Training Domain=ToolUse2026.05 | 43.97 | |
| SDPOModel Size=Qwen3-4B, Training Domain=Physics2026.05 | 43.63 | |
| GRPOModel Size=Qwen3-4B, Training Domain=Physics2026.05 | 43.6 | |
| ROSDModel Size=Qwen3-4B, Training Domain=Physics2026.05 | 43.43 | |
| GRPOModel Size=Qwen3-4B, Training Domain=Material2026.05 | 42.88 | |
| GRPOModel Size=Qwen3-8B, Training Domain=ToolUse2026.05 | 42.47 | |
| ROSDModel Size=Qwen3-8B, Training Domain=Material2026.05 | 42.12 | |
| ROSDModel Size=Qwen3-8B, Training Domain=Physics2026.05 | 41.62 | |
| GRPOModel Size=Qwen3-8B, Training Domain=Biology2026.05 | 41.58 | |
| ROSDModel Size=Qwen3-4B, Training Domain=Biology2026.05 | 40.93 | |
| SDPOModel Size=Qwen3-4B, Training Domain=Material2026.05 | 40.03 | |
| SDPOModel Size=Qwen3-4B, Training Domain=Biology2026.05 | 39.35 | |
| ROSDModel Size=Qwen3-8B, Training Domain=Biology2026.05 | 38.21 | |
| SDPOModel Size=Qwen3-8B, Training Domain=Biology2026.05 | 34.78 | |
| SDPOModel Size=Qwen3-4B, Training Domain=ToolUse2026.05 | 0.83 |