Tool-calling on AppWorld (test-challenge)
9.1Task Goal Completion (TGC)Memp
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MempModel=Qwen3-32B2026.06 | 9.1 | 1.4 | |
| ReAct + STModel=Qwen3-32B2026.06 | 8.6 | 0 | |
| KATEModel=Qwen3-32B2026.06 | 7.4 | 0.7 | |
| ReActModel=Qwen3-32B2026.06 | 6.2 | 1.4 | |
| KATEModel=Qwen3-8B2026.06 | 5.5 | 0.7 | |
| ReAct + STModel=Qwen3-8B2026.06 | 4.8 | 0.7 | |
| ReActModel=Qwen3-8B2026.06 | 3.8 | 0.7 | |
| MempModel=Qwen3-8B2026.06 | 3.6 | 0 |