https://github.com/dongjunlee/vae-tensorflow
TensorFlow implementation of Auto-Encoding Variational Bayes.
https://github.com/dongjunlee/vae-tensorflow
generative-model hb-experiment mnist tensorflow variational-autoencoders variational-inference
Last synced: 3 months ago
JSON representation
TensorFlow implementation of Auto-Encoding Variational Bayes.
- Host: GitHub
- URL: https://github.com/dongjunlee/vae-tensorflow
- Owner: DongjunLee
- Created: 2018-01-30T11:51:56.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2018-02-03T09:30:03.000Z (over 8 years ago)
- Last Synced: 2025-04-15T00:13:40.189Z (about 1 year ago)
- Topics: generative-model, hb-experiment, mnist, tensorflow, variational-autoencoders, variational-inference
- Language: Python
- Homepage:
- Size: 679 KB
- Stars: 8
- Watchers: 2
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Variational Autoencoder [](https://github.com/hb-research)
TensorFlow implementation of [Auto-Encoding Variational Bayes](https://arxiv.org/abs/1312.6114).
## Requirements
- Python 3.6
- TensorFlow >= 1.4
- [hb-config](https://github.com/hb-research/hb-config) (Singleton Config)
- requests
- [Slack Incoming Webhook URL](https://my.slack.com/services/new/incoming-webhook/)
- Matplotlib
## Project Structure
init Project by [hb-base](https://github.com/hb-research/hb-base)
.
├── config # Config files (.yml, .json) using with hb-config
├── data # dataset path
├── variational_autoencoder # VAE architecture graphs (from input to logits)
└── __init__.py # Graph logic
├── data_loader.py # download data -> generate_batch (using Dataset)
├── main.py # define experiment_fn
└── model.py # define EstimatorSpec
Reference : [hb-config](https://github.com/hb-research/hb-config), [Dataset](https://www.tensorflow.org/api_docs/python/tf/data/Dataset#from_generator), [experiments_fn](https://www.tensorflow.org/api_docs/python/tf/contrib/learn/Experiment), [EstimatorSpec](https://www.tensorflow.org/api_docs/python/tf/estimator/EstimatorSpec)
## Config
Can control all **Experimental environment**.
example: mnist.yml
```yml
model:
batch_size: 32
z_dim: 20
n_output: 784
encoder_h1: 512
encoder_h2: 256
encoder_h3: 128
decoder_h1: 128
decoder_h2: 256
decoder_h3: 512
train:
learning_rate: 0.00001
optimizer: 'Adam' # Adagrad, Adam, Ftrl, Momentum, RMSProp, SGD
train_steps: 200000
model_dir: 'logs/mnist'
save_checkpoints_steps: 1000
check_hook_n_iter: 1000
min_eval_frequency: 10
print_verbose: True
debug: False
slack:
webhook_url: "" # after training notify you using slack-webhook
```
* debug mode : using [tfdbg](https://www.tensorflow.org/programmers_guide/debugger)
## Usage
Install requirements.
```pip install -r requirements.txt```
Then, start training model
```
python main.py --config mnist
```
After training, generate image from latent vector.
```
python generate.py --config mnist --batch_size 100
```
### Experiments modes
:white_check_mark: : Working
:white_medium_small_square: : Not tested yet.
- :white_check_mark: `evaluate` : Evaluate on the evaluation data.
- :white_medium_small_square: `extend_train_hooks` : Extends the hooks for training.
- :white_medium_small_square: `reset_export_strategies` : Resets the export strategies with the new_export_strategies.
- :white_medium_small_square: `run_std_server` : Starts a TensorFlow server and joins the serving thread.
- :white_medium_small_square: `test` : Tests training, evaluating and exporting the estimator for a single step.
- :white_check_mark: `train` : Fit the estimator using the training data.
- :white_check_mark: `train_and_evaluate` : Interleaves training and evaluation.
---
### Tensorboar
```tensorboard --logdir logs```
## Result
- Generate Mnist image (Config: `mnist.yml`)
## Reference
- [hb-research/notes - Auto-Encoding Variational Bayes](https://github.com/hb-research/notes/blob/master/notes/vae.md)
- [Paper - Auto-Encoding Variational Bayes](https://arxiv.org/abs/1609.05473)
- [shaohua0116/VAE-Tensorflow ](https://github.com/shaohua0116/VAE-Tensorflow)
## Author
[Dongjun Lee](https://github.com/DongjunLee) (humanbrain.djlee@gmail.com)