Efficient Table QA via TableGrid Navigation and Progressive Inference Prompting
Open paper
Benchmark
Model
Rank
Results
table-question-answering-on-tablebench
CoT (Qwen3-8B)
–
Score: 41.69
table-question-answering-on-tablebench
ReAct (Qwen3-8B)
–
Score: 47.17
table-question-answering-on-tablebench
TGN (Qwen3-8B)
–
Score: 48.46
Rank counts only results with a code link.