HAS-QA: Hierarchical Answer Spans Model for Open-Domain Question Answering
Authors: Liang Pang, Yanyan Lan, Jiafeng Guo, Jun Xu, Lixin Su, Xueqi Cheng6875-6882
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | experiments on public Open QA datasets show that it significantly outperforms traditional RC baselines and recent Open QA baselines. |
| Researcher Affiliation | Academia | CAS Key Laboratory of Network Data Science and Technology, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China University of Chinese Academy of Sciences, Beijing, China Department of Statistics, University of California, Berkeley |
| Pseudocode | Yes | Algorithm 1 HAS-QA Model in Training Phase; Algorithm 2 HAS-QA Model in Inference Phase |
| Open Source Code | Yes | The code will be released at https://gitlab.com/pl8787/has-qa. |
| Open Datasets | Yes | We evaluate our model on three Open QA datasets, Quasar T (Dhingra, Mazaitis, and Cohen 2017), Trivia QA (Joshi et al. 2017) and Search QA (Dunn et al. 2017). |
| Dataset Splits | Yes | Quasar T and Search QA have official develop set and test set, while Trivia QA s test set is unknown, thus we split a develop set from train set and evaluate on official develop set. |
| Hardware Specification | No | The paper does not provide specific details on the hardware used for running the experiments (e.g., GPU models, CPU types, or memory specifications). |
| Software Dependencies | No | The paper mentions using 'Adadelta optimizer (Zeiler 2012)' and 'Glo Ve (Pennington, Socher, and Manning 2014)' but does not specify version numbers for other ancillary software dependencies like programming languages (e.g., Python), deep learning frameworks (e.g., TensorFlow, PyTorch), or specific libraries. |
| Experiment Setup | Yes | In training step, we use the Adadelta optimizer (Zeiler 2012) with the batch size of 30, and we choose the model performed the best on develop set. The hidden dimension of GRU is 200, and the dropout ratio is 0.8. [...] The parameters of beam search are K1 = 3 and K2 = 1. |