Code Generation on LiveCodeBench v5 (Pass@16)
45.34Pass@16TAO-RL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TAO-RLModel scale=7B2026.06 | 45.34 | 1,971.88 | 28.27 | |
| SimpleTIRModel scale=7B2026.06 | 43.86 | 1,733.98 | 28.08 | |
| AEPOModel scale=7B2026.06 | 43.41 | 1,697.81 | 19.95 | |
| TIRModel scale=7B2026.06 | 25.91 | 302.45 | 4.11 | |
| Base ModelModel scale=7B2026.06 | 14.2 | 1,653.78 | 2.76 |