2WikiMultihopQA

The '2WikiMultihopQA' dataset is a benchmark that contains multi-hop question-answer pairs derived from Wikipedia, used to evaluate the ability of models to perform reasoning across multiple documents to answer complex questions.

Papers using 2WikiMultihopQA (6)

Cascading Hallucination in Agentic RAG: The CHARM Framework for Detection and Mitigation2026Retrieval as Reasoning: Self-Evolving Agent-Native Retrieval via LLM-Wiki2026Critic-R: Improving Agentic Search using Instruction-tuned Retrievers with Natural Language Introspective Feedback2026GRASP: Graph Agentic Search over Propositions for Multi-hop Question Answering2026MiniPIC: Flexible Position-Independent Caching in <100LOC2026A2RAG: Adaptive Agentic Graph Retrieval for Cost-Aware and Reliable Reasoning2026