Yoctol/text-autoencoder

Name: text-autoencoder

Owner: YOCTOL INFO INC.

Description: Various autoencoder for text data.

Created: 2018-04-17 08:24:33.0

Updated: 2018-05-08 09:47:23.0

Pushed: 2018-05-08 09:47:26.0

Homepage: null

Size: 33

Language: Python

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

text-autoencoder

Various autoencoder for text data.

Usage

Grab one autoencoder first

 text_autoencoder.variational_autoencoders import VAEXXX
l = VAEXXX(n_steps=..., latent_size=..., state_size=..., ...)

How to train

• Warning: please preprocess your data to be a numpy array with shape (data_size, maxlen, embedding_size)

  l.fit(x=..., mask=..., epochs=10)
  

How to save model

l.save(output_path)

How to get latent vector z

l.get_latent_vector(x=..., mask=..., batch_size=1)

How to get output of encoder

l.encode(x=..., mask=..., batch_size=1)

How to load a trained model

l.load(path)

How to monitor the training process

• get the output_dir you input when calling model.fit

• monitor training loss

  nsorboard --logdir="<output_dir>/summary/subtrain/"
  

• monitor validation loss

  nsorboard --logdir="<output_dir>/summary/valid/"