Causal Discovery on Asia (d=8, |E|=8) small-scale (test)
100PrecisionAVICI (Baseline)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| AVICI (Baseline)Base Algorithm=AVICI2026.01 | 100 | 57.5 | 72.8 | 3.4 | |
| CauScientist (Qwen3-32B)Base Algorithm=AVICI, LLM Backbone=Qwen3-32B2026.01 | 100 | 95 | 97.3 | 0.4 | |
| CauScientist (Qwen3-14B)Base Algorithm=AVICI, LLM Backbone=Qwen3-14B2026.01 | 97.8 | 92.5 | 94.6 | 0.8 | |
| Qwen3-32BEvaluation Protocol=Zero-shot, Base Algorithm=Pure LLM2026.01 | 90.6 | 92.5 | 91.5 | 1.4 | |
| CauScientist (Qwen3-32B)Base Algorithm=FCI, LLM Backbone=Qwen3-32B2026.01 | 83.1 | 92.5 | 87.5 | 2.2 | |
| CauScientist (Qwen3-14B)Base Algorithm=FCI, LLM Backbone=Qwen3-14B2026.01 | 76.5 | 87.5 | 81.3 | 3.2 | |
| Qwen3-14BEvaluation Protocol=Zero-shot, Base Algorithm=Pure LLM2026.01 | 67.4 | 82.5 | 73.7 | 5 | |
| FCI (Baseline)Base Algorithm=FCI2026.01 | 66.7 | 25 | 36.4 | 6 |