Aversiveness Reduction Analysis on Phenom Aversive Tool-Use Dataset reset state
1.52Delta (Δ)Gemini 2.5 Pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 2.5 Pron=292026.03 | 1.52 | 8.7 | 1.22 | |
| Opus 4.5n=312026.03 | 1.48 | 1 | 2.96 | |
| Grok 4n=1592026.03 | 1.38 | 1 | 0.6 | |
| Overalln=2832026.03 | 1.17 | 4.6 | 0.62 | |
| GPT-5.1n=642026.03 | 0.38 | 1.1 | 0.39 |