Tabular Reasoning on Tabular Reasoning
74.31F1 ScoreSCOPE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SCOPEBackbone=GPT-OSS2026.04 | 74.31 | 62.48 | 78.27 | |
| CoTBackbone=GPT-OSS2026.04 | 74.17 | 61.77 | 78.05 | |
| Few-ShotBackbone=GPT-OSS2026.04 | 73.99 | 61.55 | 77.85 | |
| Zero ShotBackbone=GPT-OSS2026.04 | 73.5 | 61.05 | 77.47 | |
| CoTBackbone=Llama-3.32026.04 | 70.87 | 57.83 | 75.15 | |
| SCOPEBackbone=Llama-3.32026.04 | 70.87 | 60.66 | 76.12 | |
| Few-ShotBackbone=Llama-3.32026.04 | 69.38 | 56.56 | 74.05 | |
| Zero ShotBackbone=Llama-3.32026.04 | 66.96 | 54.55 | 72.04 | |
| SCOPEBackbone=Qwen32026.04 | 63.19 | 52.07 | 69.45 | |
| Zero ShotBackbone=Qwen32026.04 | 56.32 | 44.95 | 62.73 | |
| CoTBackbone=Qwen32026.04 | 55.37 | 44.65 | 61.93 | |
| Few-ShotBackbone=Qwen32026.04 | 54.74 | 44.09 | 61.48 | |
| TableGPT2Backbone=TableGPT2 architecture2026.04 | 44.03 | 33.78 | 50.81 | |
| EHRAgentBackbone=GPT-OSS2026.04 | 34.85 | 31.23 | 35.65 | |
| EHRAgentBackbone=Qwen32026.04 | 32.99 | 29.79 | 33.74 | |
| EHRAgentBackbone=Llama-3.32026.04 | 30.99 | 28.07 | 31.69 | |
| BlendSQLBackbone=Qwen32026.04 | 11.56 | 6.52 | 20.63 | |
| BlendSQLBackbone=GPT-OSS2026.04 | 7.3 | 6.48 | 7.71 | |
| BlendSQLBackbone=Llama-3.32026.04 | 5.6 | 5.15 | 5.81 |