Joint Tagging on Amazon, Book, Teams, and Sushi domains
96CAST-TCAST
Evaluation results
| Method | Links | |
|---|---|---|
| CASTBackbone=GPT-5.22026.01 | 96 | |
| TbSBackbone=GPT-5.22026.01 | 95.5 | |
| APBackbone=GPT-5.22026.01 | 95 | |
| BaselineBackbone=GPT-5.22026.01 | 94 | |
| Few-shotBackbone=GPT-5.22026.01 | 93.1 | |
| Self-ConsistencyBackbone=GPT-5.22026.01 | 91.6 | |
| CASTBackbone=DeepSeek-V3.22026.01 | 91.4 | |
| TbSBackbone=DeepSeek-V3.22026.01 | 90.4 | |
| Few-shotBackbone=DeepSeek-V3.22026.01 | 89.3 | |
| APBackbone=DeepSeek-V3.22026.01 | 89 | |
| Self-ConsistencyBackbone=DeepSeek-V3.22026.01 | 87.9 | |
| BaselineBackbone=DeepSeek-V3.22026.01 | 87.8 | |
| TbSBackbone=Gemini-3-Flash2026.01 | 86 | |
| Few-shotBackbone=Gemini-3-Flash2026.01 | 83.2 | |
| CASTBackbone=Gemini-3-Flash2026.01 | 82.6 | |
| Self-ConsistencyBackbone=Gemini-3-Flash2026.01 | 82.2 | |
| BaselineBackbone=Gemini-3-Flash2026.01 | 81.8 | |
| APBackbone=Gemini-3-Flash2026.01 | 79.3 |