CRAFT: A Unified Counterfactual Reasoning Framework for Tabular Question Answering and Fact Verification
Researchers have introduced CRAFT, a novel framework designed to enhance large language models' (LLMs) ability to reason over tabular data. CRAFT employs a bidirectional verification process, generating both declarative statements and their counterfactual alternatives to improve inference. This approach has demonstrated superior performance on table reasoning benchmarks like WikiTQ and TabFact, particularly for complex question-answering tasks. AI
IMPACT Enhances LLM capabilities in structured data reasoning, potentially improving applications requiring complex table analysis.