Exploratory Optimization on MCP-Universe
45.52Repository Management Success RateReAct + TPGO
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ReAct + TPGOBackbone=DeepSeek-V3.2, Iterations=52026.04 | 45.52 | 63.16 | 28.03 | 30.73 | 41.86 | |
| ReAct + TPGOBackbone=DeepSeek-V3.2, Iterations=32026.04 | 44.18 | 63.16 | 26.13 | 21.81 | 38.82 | |
| ReActBackbone=DeepSeek-V3.2, Iterations=02026.04 | 42.18 | 60.03 | 20.51 | 16.36 | 34.77 | |
| ReAct + TPGOBackbone=GPT-4.1, Iterations=32026.04 | 41.21 | 60.53 | 32.41 | 12.73 | 36.72 | |
| ReAct + TPGOBackbone=GPT-4.1, Iterations=52026.04 | 41.21 | 65.79 | 33.74 | 14.55 | 38.82 | |
| ReAct + TPGOBackbone=DeepSeek-V3.2, Iterations=12026.04 | 40.85 | 60.52 | 20.38 | 17.14 | 34.72 | |
| ReAct + TPGOBackbone=GPT-4.1, Iterations=12026.04 | 40.79 | 52.63 | 27.05 | 8.33 | 32.2 | |
| ReActBackbone=GPT-4.1, Iterations=02026.04 | 36.7 | 55.26 | 26.41 | 5.45 | 30.96 |