Mathematical Reasoning on AIME24 (pass@1, pass@128)
8Pass@1GRPO (Countdown)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GRPO (Countdown)Backbone=Qwen2.5-7B, Generation budget=4096 tokens, Optimization=GRPO2026.04 | 8 | 30 | |
| BaseBackbone=Qwen2.5-7B, Generation budget=4096 tokens2026.04 | 7 | 27 | |
| Countdown (A')Backbone=Qwen2.5-7B, Generation budget=4096 tokens, Training Data=Self-revised answers (A')2026.04 | 7 | 30 | |
| Countdown (A') + GRPOBackbone=Qwen2.5-7B, Generation budget=4096 tokens, Training Data=Self-revised answers (A'), Optimization=GRPO2026.04 | 7 | 40 | |
| Countdown (A) + GRPOBackbone=Qwen2.5-7B, Generation budget=4096 tokens, Training Data=LLaMA-70B teacher answers (A), Optimization=GRPO2026.04 | 6 | 23 | |
| Countdown (A)Backbone=Qwen2.5-7B, Generation budget=4096 tokens, Training Data=LLaMA-70B teacher answers (A)2026.04 | 5 | 27 |