True/False Question Answering on TMQA
0.7675AccuracyCADE
Evaluation results
| Method | Links | |
|---|---|---|
| CADEmodel size=0.6B, backbone=encoder-based2026.06 | 0.7675 | |
| T3LLMworker_llm=DeepSeek-R1, student_llm=Qwen2.5-14B-Instruct, fine_tune_method=LoRA, lora_rank=64, lora_alpha=1282025.12 | 0.766 | |
| DeepSeek-V3.2mode=zero-shot, serialization=text2026.06 | 0.7475 | |
| ChatTS2025.12 | 0.637 | |
| TSCoT2025.12 | 0.635 | |
| TSCoT+2025.12 | 0.629 | |
| ChatTime2025.12 | 0.619 | |
| Time-MQAimplementation=our implementation2025.12 | 0.564 | |
| TSTxt2025.12 | 0.539 | |
| TSEnc2025.12 | 0.518 |