Agent Tool-use and Reasoning on SEAL (test)
51.97Pass@3ExpSeek
Evaluation Results
| Method | Links | |
|---|---|---|
| ExpSeekBase Model=Qwen3-32B2026.01 | 51.97 | |
| ExpSeekBase Model=Qwen3-8B2026.01 | 50 | |
| REASONINGBANK+Base Model=Qwen3-32B2026.01 | 48.82 | |
| Training-Free GRPOBase Model=Qwen3-32B2026.01 | 48.35 | |
| No ExperienceBase Model=Qwen3-32B2026.01 | 47.64 | |
| REASONINGBANK+Base Model=Qwen3-8B2026.01 | 44.09 | |
| Training-Free GRPOBase Model=Qwen3-8B2026.01 | 42.13 | |
| No ExperienceBase Model=Qwen3-8B2026.01 | 39.76 |