Legal Reasoning on LegalBench Hearsay
86.46AccuracySciDC (Qwen3-14B)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SciDC (Qwen3-14B)Backbone=Qwen3-14B, Framework=SciDC2026.04 | 86.46 | — | — | — | |
| Claude Sonnet 3.5Backbone=Claude Sonnet 3.52026.04 | 86.17 | — | — | — | |
| GPT-4Backbone=GPT-42026.04 | 85.11 | — | — | — | |
| SciDC (Qwen3-8B)Backbone=Qwen3-8B, Framework=SciDC2026.04 | 78.72 | — | — | — | |
| Qwen3-14BBackbone=Qwen3-14B2026.04 | 78.26 | — | — | — | |
| MADBackbone=Gemini-2.5-Flash2026.02 | 77.6 | 78.5 | 76.1 | 76.5 | |
| CoTBackbone=Gemini-2.5-Flash2026.02 | 77.2 | 77.4 | 75.6 | 75.9 | |
| ACALBackbone=Gemini-2.5-Flash2026.02 | 76.7 | 77.1 | 77.6 | 76.7 | |
| SPBackbone=Gemini-2.5-Flash2026.02 | 75.5 | 75.7 | 74.2 | 74.5 | |
| RAGBackbone=Gemini-2.5-Flash2026.02 | 75.5 | 79.5 | 72.8 | 73 | |
| MADBackbone=Gemini-2.5-Flash-Lite2026.02 | 74.5 | 75.8 | 72.4 | 72.7 | |
| ACALBackbone=Gemini-2.5-Flash-Lite2026.02 | 74.5 | 77.1 | 76.3 | 74.4 | |
| RAGBackbone=Gemini-2.5-Flash-Lite2026.02 | 71.3 | 70.8 | 70.9 | 70.9 | |
| SPBackbone=Gemini-2.5-Flash-Lite2026.02 | 69.2 | 73.6 | 71.5 | 68.9 | |
| CoTBackbone=Gemini-2.5-Flash-Lite2026.02 | 69.2 | 74.7 | 71.8 | 68.7 | |
| Qwen3-8BBackbone=Qwen3-8B2026.04 | 54.84 | — | — | — |