Text Classification on Retweets (full)
66.5AccuracyRLIE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RLIEBackbone=Qwen3-235B2025.10 | 66.5 | 66.5 | |
| RLIEBackbone=DeepSeek-V3.22025.10 | 65.7 | 65.6 | |
| Zero-shotBackbone=DeepSeek-V3.22025.10 | 63.5 | 62.6 | |
| HypoGeniCBackbone=DeepSeek-V3.22025.10 | 61.9 | 61.8 | |
| RLIEBackbone=Qwen3-Next-80B2025.10 | 61.9 | 61.8 | |
| Zero-shot GenBackbone=DeepSeek-V3.22025.10 | 58.1 | 58.1 | |
| Few-shot (ICL)Backbone=DeepSeek-V3.22025.10 | 57.7 | 53.9 | |
| IO RefinementBackbone=DeepSeek-V3.22025.10 | 57.1 | 56.1 |