https://github.com/johnpertoft/relation-networks

Tensorflow implementation of Relation Networks
https://github.com/johnpertoft/relation-networks

relation-network relation-networks tensorflow

Last synced: about 1 month ago
JSON representation

Tensorflow implementation of Relation Networks

• Host: GitHub
• URL: https://github.com/johnpertoft/relation-networks
• Owner: johnPertoft
• Created: 2018-08-01T14:32:45.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2018-10-05T07:20:34.000Z (over 7 years ago)
• Last Synced: 2025-01-02T18:23:43.813Z (over 1 year ago)
• Topics: relation-network, relation-networks, tensorflow
• Language: Python
• Homepage:
• Size: 172 KB
• Stars: 4
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Relation Networks

This is a Tensorflow implementation of 

[A simple neural network module for relational reasoning](https://arxiv.org/abs/1706.01427).

Currently only the simple *Sort-of-CLEVR* dataset (described in paper) is implemented. 

This dataset simplifies the problem by considering a smaller space of questions that can

be encoded as binary strings and thus bypassing the need for a model to parse 

natural language but still retaining the need for relational reasoning.

## Sort-of-CLEVR

```bash

$ python -m rn.data.sort_of_clevr.inspect

```

![soft-of-clevr-sample](pics/sort-of-clevr-sample.png)

## Training

```bash

$ python -m rn.train

```

## Results

```bash

$ python -m rn.predict --checkpoint path/to/checkpoint [--only-misclassifications] [--seed xxx]

```

![sort-of-clevr-prediction-sample](pics/sort-of-clevr-prediction-sample.png)

TODO: Show accuracy and loss curves.