Linfeng Song
Linfeng Song
请问能不能给出一个预测predicate的system？
Hi, The data is left at home while I'm traveling right now. I'll upload the model as well once I returned home.
The model should be downloadable now.
I the code was developed under Python 2, and I'm not very familiar with Python 3. Is it possible to just stick to Python 2?
Hi, Sorry, I'm unable to make the change. You can try our most recent model https://github.com/Soistesimmer/AMR-multiview, which performs better. But you have to train by yourself.
Hi, How did you preprocess the data? You need to simplify the AMR side and tokenize the English side with a PTB_tokenizer.
I think the simplifier does tokenization for you, and you could use your own tokenizer too. If you directly use my released model for decoding, then you have to use...
Can you share some of your processed data? Also I use pretrained embeddings from Glove 840B.
It's obvious that you didn't tokenize your sentences. You can try using https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/tokenizer_PTB.perl, and then you have to re-do everyting, such as extracting vocabulary, extracting embedding, making JSON files for...
Hope that can help.