an implementation of emojify by Keras,pyTorch version to pyTorch emojify
- Coursera course by Andrew Ng
glove.6B.50d.txt is word embeding we used in the code,and it is already trained in a very large datasets by glove. It transforms every word index into 50 dimentional embeding vector.You may know glove and word2vec are both common way to do word embeding.
I have uploaded this to Google Drive yet,download and put it into data subdirectory.
Pretty simple and common model,which is very useful for emojify and some tasks like this,say many-to-one model.
datasets like this,check it in data directory.I have put all test data into train data,so you dont need to test. Because I want more data to get model better generated.
Not bad!
In fact,you are supposed to train it yourself,only takes a few minites to train about 100 epochs on GPU. However,I also upload my checkpoints on Google Drive.
Feel free to download,and put into checkpoints subdirectory,then try to run demo.py with my checkpoint,you can change your test word in demo,but notice that it should not over 10 words.