Human Process Distribution Matching on Iowa Gambling Task (IGT)
0.42Mean Absolute Cohen's dCentaur
Evaluation Results
| Method | Links | |
|---|---|---|
| CentaurParameter Scale=70B2026.05 | 0.42 | |
| Gemini 2.5 Pro2026.05 | 0.56 | |
| Base QwenParameter Scale=1.5B2026.05 | 0.81 | |
| GPT-52026.05 | 1.91 | |
| Claude Sonnet2026.05 | 3.32 |