run preprocess.sh to generate word_glove.npy about absa_keras HOT 3 OPEN

Hi, I had a small question. If I wanted to test this model for a single sentence, which script would I need to run? Because the repository has only a train.py file, so every time I run that it begins training from scratch. Is there a particular script that I could run to test the model on a single sentence?
I know your time is extremely valuable, so don't feel the need to respond in depth. However, if you do have 5 minutes, I would really appreciate it. @AlexYangLi

from absa_keras.

Hi, I had a small question. If I wanted to test this model for a single sentence, which script would I need to run? Because the repository has only a train.py file, so every time I run that it begins training from scratch. Is there a particular script that I could run to test the model on a single sentence?
I know your time is extremely valuable, so don't feel the need to respond in depth. However, if you do have 5 minutes, I would really appreciate it. @AlexYangLi

Basically , you would need to write a script that loads the model , does the preprocessing (according to the model) to do the prediction. I am in the midsts of doing a single prediction method for it. It will be in my fork once I've done the work (I'm working on other things)

from absa_keras.

@rahil1304 - You can check out my fork of this repo - there is a predict.py. It does prediction at sentence level (right now) - but you can do it at each individual word level

Basically you put in your single sentence as csv , it will generate the embeddings , then it loads the model and it would do a test against it. It's super dirty and quick hack. I'll make it bit more better

from absa_keras.