Multiple Choice Question Answering on Time-MQA
58.5AccuracyDeepSeek-V3.2
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeek-V3.2mode=zero-shot, serialization=text2026.06 | 58.5 | |
| CADEmodel size=0.6B, backbone=encoder-based2026.06 | 53.15 | |
| CADE2026.06 | 53.15 | |
| CADE w/o SupCon2026.06 | 51.89 | |
| Frozen Time-MoE2026.06 | 51.13 | |
| ITFormer2026.06 | 50.13 | |
| Frozen Random Linear2026.06 | 50.13 | |
| Time-MQA2026.06 | 47.1 | |
| Time-MQA (Full FT)2026.06 | 43.32 |