Text Classification on Headlines (full)
67AccuracyRLIE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RLIEBackbone=DeepSeek-V3.22025.10 | 67 | 67 | |
| Few-shot (ICL)Backbone=DeepSeek-V3.22025.10 | 62 | 61.8 | |
| IO RefinementBackbone=DeepSeek-V3.22025.10 | 62 | 61.1 | |
| Zero-shotBackbone=DeepSeek-V3.22025.10 | 61.2 | 59.6 | |
| RLIEBackbone=Qwen3-Next-80B2025.10 | 61.1 | 60.9 | |
| RLIEBackbone=Qwen3-235B2025.10 | 60.6 | 60.4 | |
| HypoGeniCBackbone=DeepSeek-V3.22025.10 | 59.9 | 60.1 | |
| Zero-shot GenBackbone=DeepSeek-V3.22025.10 | 57.5 | 57.4 |