Corpus ID: 215768725

A Simple Yet Strong Pipeline for HotpotQA

@article{Groeneveld2020ASY,
  title={A Simple Yet Strong Pipeline for HotpotQA},
  author={Dirk Groeneveld and Tushar Khot and Mausam and A. Sabharwal},
  journal={ArXiv},
  year={2020},
  volume={abs/2004.06753}
}
  • Dirk Groeneveld, Tushar Khot, +1 author A. Sabharwal
  • Published 2020
  • Computer Science
  • ArXiv
  • State-of-the-art models for multi-hop question answering typically augment large-scale language models like BERT with additional, intuitively useful capabilities such as named entity recognition, graph-based reasoning, and question decomposition. However, does their strong performance on popular multi-hop datasets really justify this added design complexity? Our results suggest that the answer may be no, because even our simple pipeline based on BERT, named Quark, performs surprisingly well… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    Longformer: The Long-Document Transformer
    • 52
    • Open Access

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 15 REFERENCES
    BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
    • 9,746
    • Highly Influential
    • Open Access
    SQuAD: 100, 000+ Questions for Machine Comprehension of Text
    • 2,125
    • Highly Influential
    • Open Access
    HuggingFace's Transformers: State-of-the-art Natural Language Processing
    • 687
    • Open Access
    RoBERTa: A Robustly Optimized BERT Pretraining Approach
    • 1,368
    • Highly Influential
    • Open Access
    HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering
    • 270
    • Highly Influential
    • Open Access
    Constructing Datasets for Multi-hop Reading Comprehension Across Documents
    • 189
    • Open Access
    The Web as a Knowledge-base for Answering Complex Questions
    • 111
    • Open Access
    Compositional Questions Do Not Necessitate Multi-hop Reasoning
    • 44
    • Highly Influential
    • Open Access
    Multi-hop Reading Comprehension through Question Decomposition and Rescoring
    • 37
    • Highly Influential
    • Open Access
    Dynamically Fused Graph Network for Multi-hop Reasoning
    • 39
    • Open Access