Conversational Question Answering on QuAC 1,000 1
59.4AccuracySINKTRACK
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SINKTRACKModel=Llama3.1-8B-Instruct, Prompting Strategy=SINKTRACK2026.04 | 59.4 | 54.05 | |
| CoTModel=Llama3.1-8B-Instruct, Prompting Strategy=CoT2026.04 | 54.9 | 46.09 | |
| DirectModel=Llama3.1-8B-Instruct, Prompting Strategy=Direct2026.04 | 54.1 | 48.87 | |
| SINKTRACKModel=MiniCPM3-4B, Prompting Strategy=SINKTRACK2026.04 | 53.57 | 52.82 | |
| SINKTRACKModel=Qwen2.5-7B-Instruct, Prompting Strategy=SINKTRACK2026.04 | 52.93 | 47.52 | |
| DirectModel=MiniCPM3-4B, Prompting Strategy=Direct2026.04 | 52.82 | 52.79 | |
| DirectModel=Qwen2.5-7B-Instruct, Prompting Strategy=Direct2026.04 | 52.57 | 47.06 | |
| CoTModel=MiniCPM3-4B, Prompting Strategy=CoT2026.04 | 42.69 | 45.45 | |
| CoTModel=Qwen2.5-7B-Instruct, Prompting Strategy=CoT2026.04 | 38.47 | 42.11 |