Summarization on Amazon, Book, Teams, and Sushi domains
9.93CAST-SCAST
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CASTBackbone=Gemini-3-Flash2026.01 | 9.93 | 6.6 | |
| Few-shotBackbone=Gemini-3-Flash2026.01 | 9.9 | 5.56 | |
| TbSBackbone=Gemini-3-Flash2026.01 | 9.88 | 6.75 | |
| BaselineBackbone=Gemini-3-Flash2026.01 | 9.8 | 6.06 | |
| APBackbone=Gemini-3-Flash2026.01 | 9.73 | 5.92 | |
| Self-ConsistencyBackbone=Gemini-3-Flash2026.01 | 9.72 | 13.32 | |
| CASTBackbone=DeepSeek-V3.22026.01 | 9.47 | 28.82 | |
| TbSBackbone=DeepSeek-V3.22026.01 | 9.46 | 25.55 | |
| CASTBackbone=GPT-5.22026.01 | 9.39 | 48.11 | |
| APBackbone=GPT-5.22026.01 | 9.35 | 40.91 | |
| TbSBackbone=GPT-5.22026.01 | 9.34 | 37.04 | |
| BaselineBackbone=GPT-5.22026.01 | 9.24 | 38.16 | |
| Few-shotBackbone=GPT-5.22026.01 | 9.17 | 41.63 | |
| APBackbone=DeepSeek-V3.22026.01 | 8.97 | 26.82 | |
| Few-shotBackbone=DeepSeek-V3.22026.01 | 8.96 | 31.11 | |
| BaselineBackbone=DeepSeek-V3.22026.01 | 8.15 | 24.42 | |
| Self-ConsistencyBackbone=GPT-5.22026.01 | 7.4 | 100.11 | |
| Self-ConsistencyBackbone=DeepSeek-V3.22026.01 | 7.06 | 56.25 |