ReCO: A Large Scale Chinese Reading Comprehension Dataset on Opinion

Authors: Bingning Wang, Ting Yao, Qi Zhang, Jingfang Xu, Xiaochuan Wang9146-9153

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Current QA models that perform very well on many question answering problems, such as BERT (Devlin et al. 2018), only achieves 77% accuracy on this dataset, a large margin behind humans nearly 92% performance, indicating Re CO present a good challenge for machine reading comprehension.
Researcher Affiliation Industry Bingning Wang, Ting Yao, Qi Zhang, Jingfang Xu, Xiaochuan Wang Sogou Inc. Beijing, 100084, China {wangbingning, yaoting, qizhang, xujingfang, wxc}@sogou-inc.com
Pseudocode No The paper does not contain any pseudocode or algorithm blocks.
Open Source Code Yes The codes, dataset and leaderboard will be freely available at https://github.com/benywon/Re CO.
Open Datasets Yes The codes, dataset and leaderboard will be freely available at https://github.com/benywon/Re CO.
Dataset Splits No Finally, we obtain 280,000 training data and 20,000 testing data. The paper specifies training and testing data but does not explicitly mention a separate validation set split.
Hardware Specification Yes In all experiments we set the batch size to 48 and run on 8 Nvidia V100 GPUs.
Software Dependencies No The paper mentions software like 'sentencepiece', 'Bi DAF', 'BERT', and 'ELMO' but does not provide specific version numbers for these components or any other ancillary software dependencies.
Experiment Setup Yes In all experiments we set the batch size to 48 and run on 8 Nvidia V100 GPUs.