Mathematical Reasoning on AIME25 (Acc@16)
22Avg Score @16HEAPA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HEAPABackbone=Qwen3-8B2026.01 | 22 | — | |
| GRPO (w. PS)Backbone=Qwen3-8B2026.01 | 21.9 | — | |
| GRPOBackbone=Qwen3-8B2026.01 | 21.7 | — | |
| GRPO (w. PS)Backbone=Qwen3-4B2026.01 | 19.6 | — | |
| HEAPABackbone=Qwen3-4B2026.01 | 19.3 | — | |
| GRPOBackbone=Qwen3-4B2026.01 | 18.1 | — | |
| GRPOBackbone=Qwen3-1.7B2026.01 | 6.2 | — | |
| GRPO (w. PS)Backbone=Qwen3-1.7B2026.01 | 6.2 | — | |
| HEAPABackbone=Qwen3-1.7B2026.01 | 6.2 | — | |
| HEAPABackbone=Qwen3-0.6B2026.01 | 4.3 | — | |
| GRPO (w. PS)Backbone=Qwen3-0.6B2026.01 | 3.9 | — | |
| GRPOBackbone=Qwen3-0.6B2026.01 | 3.3 | — |