sounds

Thank you for the replay. I see the embedding dict is saved in the model subword-PWIM. SO we should put all words embedding into the dict? ( The training word...

Initalizing out-of-vocabulary word vectors to zero vectors seems the same as replacing the new word to ‘OOV’. I use this model to do paraphrase identification in Chinese short sentences(query) ,...

> 1400 is a lot, maybe there's a maximum number of features, such as 1024. I tried 600 features（600 fields），segment fault too