Event Causal Identification on EventStoryLine
100CHRDr.ECI
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Dr.ECImode=Not specified2026.04 | 100 | 50 | |
| Qwen2.5-1.5Bmode=Vanilla (original model)2026.04 | 83.54 | 52.97 | |
| MRBalancemode=Not specified2026.04 | 82.45 | 59.1 | |
| Llama3.2-1Bmode=Vanilla (original model)2026.04 | 76.43 | 55.58 | |
| Qwen2.5-1.5Bmode=Zero-shot CoT prompting2026.04 | 69.77 | 51.48 | |
| Qwen2.5-1.5Bmode=In-context learning2026.04 | 69.57 | 55.32 | |
| Qwen2.5-1.5Bmode=Fine-tuning with binary labels2026.04 | 66.67 | 56.74 | |
| Llama3.1-8Bmode=Not specified2026.04 | 60.59 | 58.97 | |
| Llama3.2-1Bmode=Zero-shot CoT prompting2026.04 | 60.43 | 50.74 | |
| Qwen3-30B-A3Bmode=Not specified2026.04 | 60.27 | 58.76 | |
| Llama3.2-1Bmode=In-context learning2026.04 | 58.42 | 46.43 | |
| Llama3.2-1Bmode=Fine-tuning with binary labels2026.04 | 56.55 | 55.68 | |
| GPT4mode=Not specified2026.04 | 53.3 | 51.4 | |
| MuTQAmode=Not specified2026.04 | 46.03 | 28.82 | |
| GPT3.5-turbomode=Not specified2026.04 | 43.43 | 57.36 | |
| Llama3.2-1Bmode=Fine-tuning using CoT trace data (Our)2026.04 | 9.14 | 63.44 | |
| Qwen2.5-1.5Bmode=Fine-tuning using CoT trace data (Our)2026.04 | 6.26 | 66 |