Calibration on HotpotQA
0.28ECENAACL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NAACLBackbone=Llama-3.1-8B-Instruct2026.01 | 0.28 | 0.778 | |
| Noise-awareBackbone=Llama-3.1-8B-Instruct2026.01 | 0.309 | 0.642 | |
| NAACLBackbone=Qwen2.5-7B-Instruct2026.01 | 0.312 | 0.735 | |
| Label-only SFTBackbone=Llama-3.1-8B-Instruct2026.01 | 0.319 | 0.711 | |
| Label-only SFTBackbone=Qwen2.5-7B-Instruct2026.01 | 0.321 | 0.699 | |
| Noise-awareBackbone=Qwen2.5-7B-Instruct2026.01 | 0.325 | 0.692 | |
| CoTBackbone=Qwen2.5-7B-Instruct2026.01 | 0.353 | 0.693 | |
| NAACLBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.359 | 0.663 | |
| EnsembleBackbone=Qwen2.5-7B-Instruct2026.01 | 0.38 | 0.681 | |
| VanillaBackbone=Qwen2.5-7B-Instruct2026.01 | 0.391 | 0.712 | |
| NAACLBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.391 | 0.702 | |
| EnsembleBackbone=Llama-3.1-8B-Instruct2026.01 | 0.397 | 0.65 | |
| Noise-awareBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.422 | 0.523 | |
| Noise-awareBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.443 | 0.633 | |
| CoTBackbone=Llama-3.1-8B-Instruct2026.01 | 0.444 | 0.645 | |
| VanillaBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.457 | 0.66 | |
| VanillaBackbone=Llama-3.1-8B-Instruct2026.01 | 0.46 | 0.605 | |
| EnsembleBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.465 | 0.65 | |
| CoTBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.484 | 0.617 | |
| VanillaBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.499 | 0.641 | |
| EnsembleBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.515 | 0.614 | |
| Label-only SFTBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.517 | 0.577 | |
| CoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.529 | 0.632 | |
| Label-only SFTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.591 | 0.538 |