Reasoning on GPQA-Diamond (generalization)
41.6AccuracyAutoTTS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AutoTTSType=Discovered, Base Model=Qwen3-1.7B, β=12026.05 | 41.6 | 270,100 | |
| AutoTTSType=Discovered, Base Model=Qwen3-1.7B, β=0.52026.05 | 41.6 | 151,000 | |
| SC@64Type=Handcrafted, Base Model=Qwen3-1.7B2026.05 | 41.3 | 510,000 | |
| ESCType=Handcrafted, Base Model=Qwen3-1.7B2026.05 | 41.3 | 391,600 | |
| ASCType=Handcrafted, Base Model=Qwen3-1.7B2026.05 | 41 | 186,300 |