Open InesArous opened 3 years ago
Hi @InesArous, thanks for your interest! I think it should be a text file with each line as one of the tags in your data. The file is used in this function for creating the vocabulary to convert tags into indices https://github.com/mingdachen/vsl/blob/efee26646c33832f2004bacf8afcb2f4d18ed942/data_utils.py#L118
Thanks a lot for your reply! I figured it is generated in the process_twitter_data.py but the naming was different. That's why I got confused. On another note, is it possible to provide the hyper-parameters that were used with the conll2003 dataset? That would be really helpful. Thanks in advance!
@InesArous Sorry about that! Regarding hyperparameters, I can't remember the hyperparameters used for that particular dataset. In case this is helpful, we have reported the ranges of hyperparameters we tuned in the appendix. See Sec. A in this file https://aclanthology.org/attachments/D18-1020.Attachment.pdf
Hello, The work is very cool. Which tag file to use? it is indicated as tagger_twitter but not included in this repo. Thanks!