Table Reasoning on TableBench
72.52Pass@1ReAct
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ReActBackbone=DeepSeek-V3.12025.09 | 72.52 | 81.68 | |
| ReActBackbone=o4-mini2025.09 | 71.03 | 80.15 | |
| DATAMINDBackbone=Qwen-2.5-Coder-14B2025.09 | 70.95 | 81.81 | |
| ReActBackbone=GPT-52025.09 | 69.93 | 78.37 | |
| ReActBackbone=DeepSeek-R12025.09 | 68.96 | 79.52 | |
| DATAMINDBackbone=Qwen-2.5-Coder-7B2025.09 | 67.6 | 79.39 | |
| ReActBackbone=Qwen-2.5-72B2025.09 | 65.44 | 76.21 | |
| ReActBackbone=GPT-4o2025.09 | 64.97 | 75.06 | |
| ReActBackbone=Qwen-2.5-Coder-32B2025.09 | 61.11 | 72.26 | |
| ReActBackbone=QwQ-32B2025.09 | 57.79 | 75.19 | |
| ReActBackbone=Qwen-2.5-Coder-14B2025.09 | 56.96 | 69.97 | |
| Table-R1Backbone=Qwen-2.5-Coder-7B2025.09 | 56.36 | 63.61 | |
| ReActBackbone=Llama-3.3-70B2025.09 | 55.47 | 70.36 | |
| Table-R1Backbone=Qwen-2.5-Coder-14B2025.09 | 50.38 | 58.91 | |
| TableLLMBackbone=Qwen-2.5-Coder-14B2025.09 | 46.44 | 76.08 | |
| OmniSQLBackbone=Qwen-2.5-Coder-14B2025.09 | 41.98 | 52.67 | |
| SQL-R1Backbone=Qwen-2.5-Coder-14B2025.09 | 41.22 | 51.02 | |
| TableLLMBackbone=Qwen-2.5-Coder-7B2025.09 | 41.01 | 70.36 | |
| SQL-R1Backbone=Qwen-2.5-Coder-7B2025.09 | 40.84 | 50.64 | |
| OmniSQLBackbone=Qwen-2.5-Coder-7B2025.09 | 39.95 | 50.25 | |
| ReActBackbone=Qwen-2.5-Coder-7B2025.09 | 11.7 | 28.63 |