Conversational Question Answering on QuAC (Acc, Macro-F1)
53.51AccuracySINKTRACK
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SINKTRACKModel=Llama3.1-8B-Instruct, Prompting Strategy=SINKTRACK2026.04 | 53.51 | 51.56 | |
| DirectModel=Llama3.1-8B-Instruct, Prompting Strategy=Direct2026.04 | 52.45 | 50.66 | |
| SINKTRACKModel=Qwen2.5-7B-Instruct, Prompting Strategy=SINKTRACK2026.04 | 50.25 | 46.73 | |
| DirectModel=Qwen2.5-7B-Instruct, Prompting Strategy=Direct2026.04 | 49.47 | 46.71 | |
| SINKTRACKModel=MiniCPM3-4B, Prompting Strategy=SINKTRACK2026.04 | 47.29 | 48.08 | |
| DirectModel=MiniCPM3-4B, Prompting Strategy=Direct2026.04 | 47.17 | 48.07 | |
| CoTModel=Llama3.1-8B-Instruct, Prompting Strategy=CoT2026.04 | 46.95 | 28.74 | |
| CoTModel=MiniCPM3-4B, Prompting Strategy=CoT2026.04 | 37.68 | 36.5 | |
| CoTModel=Qwen2.5-7B-Instruct, Prompting Strategy=CoT2026.04 | 36.48 | 36.58 |