Xml cnn

[sian-chen]

1. Implement KimCNN, XMLCNN
2. Remove get_network(), use model_name to get the network instead
3. Allow setting weight initialization method
4. Add config to log
5. Allow fixed length (for dynamic max-pooling)
6. Add config files for EUR-Lex

[cjlin1]

On 2021-04-30 16:25, sianchen wrote: Yes, you are right.. So please modify the comment to say the difference you mentioned

…

@sian-chen commented on this pull request. ------------------------- In data_utils.py [1]: > with open(config.vocab_file, 'r') as fp: - vocab_list = ['**PAD**'] + [vocab.strip() for vocab in fp.readlines()] - vocabs = Vocab(collections.Counter(vocab_list), specials=['<unk>'], + vocab_list = [PAD] + [vocab.strip() for vocab in fp.readlines()] Why we need this comment? For the other case (vocab_file is not used), PAD is also put in the beginning. -- You are receiving this because your review was requested. Reply to this email directly, view it on GitHub [1], or unsubscribe [2]. [ { ***@***.***": "http:https://schema.org", ***@***.***": "EmailMessage", "potentialAction": { ***@***.***": "ViewAction", "target": "#22 (comment)", "url": "#22 (comment)", "name": "View Pull Request" }, "description": "View this Pull Request on GitHub", "publisher": { ***@***.***": "Organization", "name": "GitHub", "url": "https://github.com" } } ] Links: ------ [1] #22 (comment) [2] https://github.com/notifications/unsubscribe-auth/ABI3BHXMCNSNQABGNGSIW5LTLJSONANCNFSM43T53ZIQ