https://github.com/maasencioh/chemicalrnn

Machine learning final project 2015-2
https://github.com/maasencioh/chemicalrnn

Last synced: about 5 hours ago
JSON representation

Machine learning final project 2015-2

• Host: GitHub
• URL: https://github.com/maasencioh/chemicalrnn
• Owner: maasencioh
• License: gpl-2.0
• Created: 2015-11-21T15:05:05.000Z (almost 9 years ago)
• Default Branch: master
• Last Pushed: 2015-12-09T16:28:34.000Z (almost 9 years ago)
• Last Synced: 2024-05-02T02:14:35.174Z (6 months ago)
• Language: Python
• Homepage:
• Size: 24.5 MB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# chemicalRNN

A SMILES random generator using Recurrent Neural Networks using Tensorflow.

### How to use it

For now you have to download [Tensorflow](https://github.com/tensorflow/tensorflow) from github and

then put the ptb_word_lm.py in the tensorflow folder: 

```{path to Tensorflow}/tensorflow/models/rnn/ptb/```

### Compile

In the path to tensorflow, you can compile as following:

``` bazel build -c opt --config=cuda tensorflow/models/rnn/ptb:ptb_word_lm ```

If you don't have cuda, you can remove this: "--config=cuda".

Then for running you can use:

``` bazel-bin/tensorflow/models/rnn/ptb/ptb_word_lm --model {size} --data_path={Path To Folder Data} 

    --load={Path to model}

```

Supported sizes:

* small

* medium

* large

* normal

for our model we use normal. The data_path must have the ptb .txt files for training, cross and test sets, and 

load must be a model file if you want to use a pre-trained model, in our case model.ckpt.

 

# Generating...

For now you have to uncomment the lines from 371 to 374, then you compile it, and run.