Table Question Answering on TableBench
62EMMATA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MATABackbone=Claude-3.7-Sonnet, Model Class=Large LLM2026.02 | 62 | 76.4 | 66.4 | |
| MixSCBackbone=Claude-3.7-Sonnet, Model Class=Large LLM2026.02 | 61.9 | 76.7 | 65.9 | |
| TabLaPBackbone=Claude-3.7-Sonnet, Model Class=Large LLM2026.02 | 61.2 | 76.3 | 65.5 | |
| MATABackbone=GPT-4o, Model Class=Large LLM2026.02 | 59.5 | 74 | 62.9 | |
| MATABackbone=cogito-32b, Model Class=Large LLM2026.02 | 57.7 | 72.3 | 60.9 | |
| MATABackbone=qwen2.5-32b, Model Class=Large LLM2026.02 | 57.7 | 72.1 | 60.7 | |
| MATABackbone=mistral-small-24b, Model Class=Large LLM2026.02 | 57.3 | 72.4 | 60.6 | |
| TabLaPBackbone=GPT-4o, Model Class=Large LLM2026.02 | 55.6 | 72.2 | 59.5 | |
| MixSCBackbone=GPT-4o, Model Class=Large LLM2026.02 | 49.4 | 69.2 | 54 | |
| SynTQABackbone=Claude-3.7-Sonnet, Model Class=Large LLM2026.02 | 48.9 | 63.3 | 54 | |
| SynTQABackbone=GPT-4o, Model Class=Large LLM2026.02 | 47.6 | 60.7 | 50.3 | |
| SynTQABackbone=cogito-32b, Model Class=Large LLM2026.02 | 44.3 | 59.1 | 48.1 | |
| TabLaPBackbone=cogito-32b, Model Class=Large LLM2026.02 | 44 | 61.4 | 48.3 | |
| MixSCBackbone=cogito-32b, Model Class=Large LLM2026.02 | 43 | 61.4 | 47.6 | |
| SynTQABackbone=qwen2.5-32b, Model Class=Large LLM2026.02 | 39.8 | 55.3 | 43.6 | |
| SynTQABackbone=mistral-small-24b, Model Class=Large LLM2026.02 | 39.1 | 54.3 | 43.1 | |
| MixSCBackbone=mistral-small-24b, Model Class=Large LLM2026.02 | 37.8 | 53 | 41 | |
| MATABackbone=llama3.2-3b, Model Class=Small LLM2026.02 | 35.4 | 56.3 | 38.1 | |
| MATABackbone=qwen2.5-7b, Model Class=Small LLM2026.02 | 35.4 | 55.7 | 39.3 | |
| TabLaPBackbone=mistral-small-24b, Model Class=Large LLM2026.02 | 32.2 | 47.8 | 35.2 | |
| SynTQABackbone=qwen2.5-7b, Model Class=Small LLM2026.02 | 30.2 | 45 | 33.6 | |
| MixSCBackbone=qwen2.5-32b, Model Class=Large LLM2026.02 | 29.7 | 55.1 | 34.1 | |
| MATABackbone=mistral-7b, Model Class=Small LLM2026.02 | 29.4 | 47.3 | 32.1 | |
| MATABackbone=qwen2.5-3b, Model Class=Small LLM2026.02 | 29.1 | 47.1 | 31.7 | |
| MATABackbone=phi4-mini-3.8b, Model Class=Small LLM2026.02 | 27.3 | 45.7 | 29.5 | |
| TabLaPBackbone=qwen2.5-32b, Model Class=Large LLM2026.02 | 26.8 | 53.3 | 31.7 | |
| SynTQABackbone=mistral-7b, Model Class=Small LLM2026.02 | 22.7 | 36.7 | 27 | |
| SynTQABackbone=qwen2.5-3b, Model Class=Small LLM2026.02 | 20.8 | 36.4 | 24.5 | |
| SynTQABackbone=phi4-mini-3.8b, Model Class=Small LLM2026.02 | 20.2 | 36.6 | 25.3 | |
| MixSCBackbone=qwen2.5-7b, Model Class=Small LLM2026.02 | 16.9 | 36.8 | 19 | |
| TabLaPBackbone=qwen2.5-3b, Model Class=Small LLM2026.02 | 16.3 | 41.7 | 19.5 | |
| MixSCBackbone=qwen2.5-3b, Model Class=Small LLM2026.02 | 16.3 | 41.7 | 19.7 | |
| MixSCBackbone=phi4-mini-3.8b, Model Class=Small LLM2026.02 | 14.4 | 41.1 | 20.3 | |
| SynTQABackbone=llama3.2-3b, Model Class=Small LLM2026.02 | 8.9 | 23.1 | 12 | |
| MixSCBackbone=mistral-7b, Model Class=Small LLM2026.02 | 8.2 | 35.5 | 15.1 | |
| MixSCBackbone=llama3.2-3b, Model Class=Small LLM2026.02 | 8.1 | 37.2 | 14.4 | |
| TabLaPBackbone=qwen2.5-7b, Model Class=Small LLM2026.02 | 7.9 | 25.5 | 9.4 | |
| TabLaPBackbone=llama3.2-3b, Model Class=Small LLM2026.02 | 6.7 | 35.7 | 13 | |
| TabLaPBackbone=phi4-mini-3.8b, Model Class=Small LLM2026.02 | 5.6 | 33.4 | 12.6 | |
| TabLaPBackbone=mistral-7b, Model Class=Small LLM2026.02 | 3.6 | 33.1 | 11.9 |