Mathematical Reasoning on HMMT-25 (average@16)
46.1Average@16Sleep
Evaluation Results
| Method | Links | |
|---|---|---|
| SleepBackbone=Qwen3-8B2026.06 | 46.1 | |
| OPSDBackbone=Qwen3-8B2026.06 | 45.1 | |
| GRPOBackbone=Qwen3-8B2026.06 | 44.9 | |
| SFTBackbone=Qwen3-8B2026.06 | 43.7 | |
| Base (Instruct)Backbone=Qwen3-8B2026.06 | 42.4 | |
| SleepBackbone=Qwen3-1.7B2026.06 | 29.3 | |
| OPSDBackbone=Qwen3-1.7B2026.06 | 28.1 | |
| GRPOBackbone=Qwen3-1.7B2026.06 | 26.1 | |
| Base (Instruct)Backbone=Qwen3-1.7B2026.06 | 25.7 | |
| SFTBackbone=Qwen3-1.7B2026.06 | 22.9 |