Question Answering on WTQ
60.9AccuracyDiVA-Former
Evaluation results
| Method | Links | |
|---|---|---|
| DiVA-FormerModality=V+T2026.03 | 60.9 | |
| DirectModality=V+T2026.03 | 55.2 | |
| DirectModality=Vision2026.03 | 55.1 | |
| DirectModality=Text2026.03 | 54.6 | |
| AdapterModality=Vision2026.03 | 52.9 | |
| AdapterModality=Text2026.03 | 52.8 | |
| AdapterModality=V+T2026.03 | 52.5 | |
| TableLlama+textLLM=Llama-2 7B, Representation=OCR textual table2026.02 | 31.63 | |
| Re-Table-7B-rerankLLM=Mistral 7B, Method Variant=rerank2026.02 | 19.19 | |
| MonkeyLLM=Qwen 7B2026.02 | 19.07 | |
| Table-LLaVA 7BLLM=Vicuna-1.5 7B2026.02 | 18.43 | |
| Re-Table-7B-retrievalLLM=Mistral 7B, Method Variant=retrieval2026.02 | 17.19 | |
| ResamplerModality=Vision2026.03 | 13.6 | |
| ResamplerModality=Text2026.03 | 13.2 | |
| ResamplerModality=V+T2026.03 | 13.2 | |
| Vary-toyLLM=Qwen 1.8B2026.02 | 7.96 | |
| Llama2+textLLM=Llama-2 7B, Representation=OCR textual table2026.02 | 4.26 | |
| BLIP2LLM=Flan-T5 7B2026.02 | 2.01 | |
| LLaVA v1.5LLM=Vicuna-1.5 7B2026.02 | 1.24 | |
| LLaVA v1.6LLM=Vicuna-1.5 7B2026.02 | 1.04 | |
| MiniGPT-4LLM=Vicuna 7B2026.02 | 0.9 |