Interpretable AMR-Based Question Decomposition for Multi-hop Question Answering

Authors: Zhenyun Deng, Yonghua Zhu, Yang Chen, Michael Witbrock, Patricia Riddle

IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on Hotpot QA demonstrate that our approach is competitive for interpretable reasoning and that the sub-questions generated by QDAMR are well-formed, outperforming existing question-decomposition-based multihop QA approaches.
Researcher Affiliation Academia School of Computer Science, University of Auckland, New Zealand {zden658, yzhu970}@aucklanduni.ac.nz, {yang.chen, m.witbrock, p.riddle}@auckland.ac.nz
Pseudocode Yes Algorithm 1 : Question Decomposition Based on AMR
Open Source Code No The paper does not provide a specific link or explicit statement about the release of its source code.
Open Datasets Yes Given ordered sub-questions, we train a single-hop QA on SQUAD [Rajpurkar et al., 2016] and on the new single-hop QA dataset. The dataset consists of single-hop QA pairs constructed by [Pan et al., 2020] on Hotpot QA.
Dataset Splits Yes Table 1: Results for QD-based multi-hop QA models on the dev set of Hotpot QA.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions software like BART, T5, Spring, RoBERTa, but does not provide specific version numbers for these or other ancillary software components.
Experiment Setup No The paper describes the general experimental process but does not provide specific hyperparameters (e.g., learning rate, batch size) or detailed system-level training settings in the main text.