Causal Discovery on Child d=20, |E|=25 medium-scale (test)
100PrecisionAVICI (Baseline)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| AVICI (Baseline)Base Algorithm=AVICI2026.01 | 100 | 24 | 38.4 | 19 | |
| CauScientist (Qwen3-32B)Base Algorithm=AVICI, LLM Backbone=Qwen3-32B2026.01 | 73.9 | 45.6 | 53.6 | 19 | |
| CauScientist (Qwen3-14B)Base Algorithm=AVICI, LLM Backbone=Qwen3-14B2026.01 | 67.7 | 52.8 | 57.4 | 18.6 | |
| Qwen3-32BEvaluation Protocol=Zero-shot, Base Algorithm=Pure LLM2026.01 | 49.7 | 55.2 | 51.5 | 25.4 | |
| Qwen3-14BEvaluation Protocol=Zero-shot, Base Algorithm=Pure LLM2026.01 | 49.3 | 46.4 | 47.6 | 23.6 | |
| CauScientist (Qwen3-32B)Base Algorithm=FCI, LLM Backbone=Qwen3-32B2026.01 | 44.9 | 17.6 | 25.2 | 21.8 | |
| CauScientist (Qwen3-14B)Base Algorithm=FCI, LLM Backbone=Qwen3-14B2026.01 | 40.6 | 20 | 26.5 | 21.6 | |
| FCI (Baseline)Base Algorithm=FCI2026.01 | 37.5 | 12 | 18.2 | 22 |