https://github.com/tomekkorbak/measuring-non-trivial-compositionality

Code accompanying the paper Measuring non-trivial compositionality in emergent communication
https://github.com/tomekkorbak/measuring-non-trivial-compositionality

compositionality deep-learning emergent-communication generalization representation-learning

Last synced: about 2 months ago
JSON representation

Code accompanying the paper Measuring non-trivial compositionality in emergent communication

• Host: GitHub
• URL: https://github.com/tomekkorbak/measuring-non-trivial-compositionality
• Owner: tomekkorbak
• Created: 2020-07-21T15:18:39.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2021-12-18T14:51:14.000Z (about 3 years ago)
• Last Synced: 2024-10-10T20:18:13.848Z (2 months ago)
• Topics: compositionality, deep-learning, emergent-communication, generalization, representation-learning
• Language: Python
• Homepage: https://arxiv.org/abs/2010.15058
• Size: 118 KB
• Stars: 6
• Watchers: 2
• Forks: 3
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Measuring non-trivial compositionality

This repo contains the source code accompanying the paper [*Measuring non-trivial compositionality in emergent communication*](https://arxiv.org/abs/2010.15058) presented at NeurIPS 2020 workshop [Talking to Strangers: Zero-Shot Emergent Communication](https://sites.google.com/view/emecom2020).

All metrics and protocols implement a common API defined in `metrics.base.Metric` and `protocols.Protocol` and were designed to be reusable.

### Running

To install all the dependencies, run `pip install -r requirements.txt`. We assume Python 3.6+.

To reproduce the results and plots from the main experiment (how each metric scores each protocol), run `python experiment_1.py`.

To reproduce the results and plots from the experiment in appendix C (how the composition function in TRE affects compositionality scores), run `python experiment_2.py`.

To run init tests for the metrics, run `pytest`.

### Citing

```bibtex

@misc{korbak2020measuring,

      title={Measuring non-trivial compositionality in emergent communication}, 

      author={Tomasz Korbak and Julian Zubek and Joanna Rączaszek-Leonardi},

      year={2020},

      eprint={2010.15058},

      archivePrefix={arXiv},

      primaryClass={cs.NE}

}

```