Principal Researcher, Huawei Technologies Ltd.
3 papers at NeurIPS 2025
MMDocRAG, a comprehensive multimodal DocRAG benchmark
LLM-as-RS paradigm is not always good: we benchmark it with traditional recommendation models