https://github.com/kefirski/pytorch_RVAE

Recurrent Variational Autoencoder that generates sequential data implemented with pytorch
https://github.com/kefirski/pytorch_RVAE

deep-learning nlp python pytorch vae

Last synced: 11 days ago
JSON representation

Recurrent Variational Autoencoder that generates sequential data implemented with pytorch

• Host: GitHub
• URL: https://github.com/kefirski/pytorch_RVAE
• Owner: kefirski
• License: mit
• Created: 2017-02-12T09:54:09.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2017-03-15T07:23:52.000Z (over 7 years ago)
• Last Synced: 2024-05-23T02:33:26.651Z (6 months ago)
• Topics: deep-learning, nlp, python, pytorch, vae
• Language: Python
• Homepage:
• Size: 1.79 MB
• Stars: 356
• Watchers: 10
• Forks: 89
• Open Issues: 12
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-pytorch - Recurrent Variational Autoencoder

README

        
# Pytorch Recurrent Variational Autoencoder 

## Model:

This is the implementation of Samuel Bowman's [Generating Sentences from a Continuous Space](https://arxiv.org/abs/1511.06349#)

with Kim's [Character-Aware Neural Language Models](https://arxiv.org/abs/1508.06615) embedding for tokens

## Sampling examples:

> the new machine could be used to increase the number of ventures block in the company 's \ shopping system to finance diversified organizations

> u.s. government officials also said they would be willing to consider whether the proposal could be used as urging and programs

> men believe they had to go on the \ because their \ were \ expensive important

> the companies insisted that the color set could be included in the program

## Usage

### Before model training it is necessary to train word embeddings:

```

$ python train_word_embeddings.py

```

This script train word embeddings defined in [Mikolov et al. Distributed Representations of Words and Phrases](https://arxiv.org/abs/1310.4546)

#### Parameters:

`--use-cuda`

`--num-iterations`

`--batch-size`

`--num-sample` –– number of sampled from noise tokens

### To train model use:

```

$ python train.py

```

#### Parameters:

`--use-cuda`

`--num-iterations`

`--batch-size`

`--learning-rate`

 

`--dropout` –– probability of units to be zeroed in decoder input

`--use-trained` –– use trained before model

### To sample data after training use:

```

$ python sample.py

```

#### Parameters:

`--use-cuda`

`--num-sample`