Calibration on StrategyQA
0.285ECENAACL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NAACLBackbone=Llama-3.1-8B-Instruct2026.01 | 0.285 | 0.624 | |
| Label-only SFTBackbone=Qwen2.5-7B-Instruct2026.01 | 0.297 | 0.679 | |
| NAACLBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.306 | 0.672 | |
| NAACLBackbone=Qwen2.5-7B-Instruct2026.01 | 0.31 | 0.726 | |
| Noise-awareBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.321 | 0.53 | |
| NAACLBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.323 | 0.651 | |
| Noise-awareBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.343 | 0.621 | |
| Label-only SFTBackbone=Llama-3.1-8B-Instruct2026.01 | 0.345 | 0.619 | |
| Label-only SFTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.347 | 0.595 | |
| CoTBackbone=Llama-3.1-8B-Instruct2026.01 | 0.354 | 0.555 | |
| CoTBackbone=Qwen2.5-7B-Instruct2026.01 | 0.363 | 0.709 | |
| EnsembleBackbone=Qwen2.5-7B-Instruct2026.01 | 0.368 | 0.719 | |
| EnsembleBackbone=Llama-3.1-8B-Instruct2026.01 | 0.37 | 0.609 | |
| Noise-awareBackbone=Llama-3.1-8B-Instruct2026.01 | 0.376 | 0.615 | |
| Noise-awareBackbone=Qwen2.5-7B-Instruct2026.01 | 0.393 | 0.618 | |
| VanillaBackbone=Llama-3.1-8B-Instruct2026.01 | 0.396 | 0.602 | |
| VanillaBackbone=Qwen2.5-7B-Instruct2026.01 | 0.398 | 0.689 | |
| EnsembleBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.399 | 0.673 | |
| Label-only SFTBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.405 | 0.554 | |
| VanillaBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.408 | 0.642 | |
| CoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.409 | 0.668 | |
| EnsembleBackbone=DeepSeek-R1-Distill-Qwen-7B2026.01 | 0.415 | 0.659 | |
| VanillaBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.416 | 0.639 | |
| CoTBackbone=DeepSeek-R1-Distill-Llama-8B2026.01 | 0.434 | 0.656 |