Investment decision alignment on P2 v1 (test)
6.12Overall MSEInvestAgent
Evaluation results
| Method | Links | ||
|---|---|---|---|
| InvestAgentBackbone=Qwen-2, Training Stage=Post-InvestAlign2025.07 | 6.12 | -60.92 | |
| InvestAgentBackbone=Llama-3.1, Training Stage=Post-InvestAlign2025.07 | 6.66 | -45.77 | |
| Llama-3.1Training Stage=Pre-SFT2025.07 | 12.28 | — | |
| Qwen-2Training Stage=Pre-SFT2025.07 | 15.66 | — |