General Chat on AE2 LC
61.7Win RateQwen3-8B (Non-thinking)
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-8B (Non-thinking)Backbone=Qwen3-8B, Training Stage=Official Post-trained2026.05 | 61.7 | |
| Qwen3-8B-GRLOBackbone=Qwen3-8B, Training Stage=GRLO2026.05 | 57.8 | |
| Qwen3-8B-GRLO+RLVRBackbone=Qwen3-8B, Training Stage=GRLO+RLVR2026.05 | 55.9 | |
| Qwen3-8B-RLVRBackbone=Qwen3-8B, Training Stage=RLVR2026.05 | 23.6 | |
| Qwen3-8B-BaseBackbone=Qwen3-8B, Training Stage=Base2026.05 | 12.8 | |
| Qwen3-8B-MathSFTBackbone=Qwen3-8B, Training Stage=MathSFT2026.05 | 7.6 |