Resource for WWW 2021 paper "Diverse and Specific Clarification Question Generation with Keywords"
Description
Training data, model outputs and pretrained checkpoints for WWW 2021 paper ``Diverse and Specific Clarification Question Generation with Keywords"
- data
- contains the processed and cleaned *Home and Kitchen* dataset
- data_office
- contains the processed and cleaned *Office* dataset
- output
- contains all outputs on the test set of *Home and Kitchen* of the 7 compared systems in Table 4
- beam: KPCNet(-filter)
- beam_filt: KPCNet(beam)
- diverse_beam: KPCNet(divbeam)
- hMup: hMup
- kwd_cluster: KPCNet(cluster)
- kwd_samples: KPCNet(sample)
- MLE: MLE
- the suffix like `beam`+X, where X indicates the order in beam search
- hMup is an exception. X indicates the id of the *expert*. `beam4` achieved the best BLEU, and is thus used in both individual and group level evaluation.
- the suffix for kwd_cluster and kwd_samples has an additional `a`+X, where X indicates the group id for conditioning keyword set
- All variants of KPCNet has additional file with suffix `kwd_samples`
- The words on the i-th line are conditioning keywords for the i-th sample
- ckpt
- pretrained model checkpoint
- hparams
- hyperparameters for the pretrained checkpoint
Files
Files
(234.6 MB)
Name | Size | Download all |
---|---|---|
md5:6d72be7053b9818c6af49a46b7f5dd79
|
67.1 MB | Download |
md5:a99c572048fdd69161ae5d9ac60f09f1
|
67.1 MB | Download |
md5:f609a564229ae48eae8e71e478d4a529
|
67.1 MB | Download |
md5:cc2ba9a6b493467166f69ac60ea4fa3b
|
33.3 MB | Download |
Additional details
Related works
- Is supplement to
- Conference paper: 10.1145/3442381.3449876 (DOI)