Supply chain management on Semi-Sim
9.15TS+TreeSearch
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| +TreeSearchStrategy=ReflAct + TreeSearch, Backbone=DeepSeek-V3.22026.06 | 9.15 | 82.12 | 4.45 | 70.12 | |
| ReflActStrategy=ReflAct, Backbone=DeepSeek-V3.22026.06 | 8.12 | 80.52 | 5.1 | 66.52 | |
| ReActStrategy=ReAct, Backbone=DeepSeek-V3.22026.06 | 7.48 | 79.03 | 5.79 | 52.33 | |
| NoThinkingStrategy=NoThinking, Backbone=DeepSeek-V3.22026.06 | 2.11 | 68.33 | 6.12 | 48.23 | |
| ReflectiChainStrategy=ReflectiChain, Backbone=DeepSeek-V3.22026.06 | 1.85 | 84.32 | 3.9 | 88.51 | |
| PPOStrategy=PPO, Backbone=DeepSeek-V3.22026.06 | -0.2 | 60.74 | 1.34 | — |