Task Generation on red-wine benchmark
61.84Utilitygpt-5.4
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| gpt-5.4generator_model=gpt-5.42026.05 | 61.84 | 60 | 40 | 4.8 | |
| gpt-5.4-minigenerator_model=gpt-5.4-mini2026.05 | 60.79 | 28 | 12 | 5.2 | |
| gpt-5.4-nano-reasoning-highgenerator_model=gpt-5.4-nano-reasoning-high2026.05 | 60.47 | 24 | 4.9 | 5.2 | |
| gpt-5.4-nano-reasoning-lowgenerator_model=gpt-5.4-nano-reasoning-low2026.05 | 60.02 | 15 | 3.6 | 5.3 | |
| gpt-5.4-nanogenerator_model=gpt-5.4-nano2026.05 | 59.8 | 12 | 3.27 | 5.5 | |
| gpt-4o-minigenerator_model=gpt-4o-mini2026.05 | 58.04 | 21 | 2 | 6.6 | |
| SH2026.05 | 56.46 | — | — | — | |
| random2026.05 | 56.02 | — | — | — | |
| hyperband2026.05 | 55.24 | — | — | — |