Mathematical Reasoning on MATH (Pass@1 & TC)
79.2Pass@1EAPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EAPOBackbone=Llama3.1-8B-Instruct2026.06 | 79.2 | 1.03 | |
| GRPOBackbone=Llama3.1-8B-Instruct2026.06 | 78.4 | 1.08 | |
| Reinforce++Backbone=Llama3.1-8B-Instruct2026.06 | 76.8 | 1.2 | |
| BaseBackbone=Llama3.1-8B-Instruct2026.06 | 62.4 | — | |
| TIRBackbone=Llama3.1-8B-Instruct2026.06 | 58 | 0.97 |