https://github.com/seba-1511/atg

Code Release for "Embedding Adaptation is Still Needed for Few-Shot Learning"
https://github.com/seba-1511/atg

Last synced: 11 days ago
JSON representation

Code Release for "Embedding Adaptation is Still Needed for Few-Shot Learning"

• Host: GitHub
• URL: https://github.com/seba-1511/atg
• Owner: seba-1511
• License: mit
• Created: 2021-01-04T18:27:06.000Z (almost 4 years ago)
• Default Branch: master
• Last Pushed: 2021-01-07T21:06:49.000Z (almost 4 years ago)
• Last Synced: 2024-12-02T18:58:31.192Z (24 days ago)
• Language: Python
• Size: 289 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • Changelog: CHANGELOG.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Embedding Adaptation is Still Needed for Few-Shot Learning

[![arXiv](https://img.shields.io/badge/arXiv-2101.XXXXX-b31b1b.svg)](https://arxiv.org/abs/2101.XXXXX)

Code Release for "Embedding Adaptation is Still Needed for Few-Shot Learning"

This code provides:

* Re-implementation of the ATG algorithm in `examples/atg.py`.

* Loaders for the dataset splits introduced in the paper.

* Demonstration code for training the algorithms, borrowed from [learn2learn](https://github.com/learnables/learn2learn).

## Resources

* Website: [seba1511.net/projects/atg](seba1511.net/projects/atg)

* Preprint: [arxiv.org/abs/2101.XXXXX](https://arxiv.org/abs/2101.XXXXX)

* Code: [github.com/Sha-Lab/atg](https://github.com/Sha-Lab/atg)

## Citation

Please cite this work as follows:

> "Embedding Adaptation is Still Needed for Few-Shot Learning", Sébastien M. R. Arnold and Fei Sha

or with the following BibTex entry:

~~~bibtex

@article{arnold2021embedding,

    title={Embedding Adaptation is Still Needed for Few-Shot Learning},

    author={Sebastien M. R. Arnold, Fei Sha},

    year={2021},

    archivePrefix={arXiv},

    primaryClass={cs.LG}

}

~~~

## Usage

Dependencies include the following Python packages:

* PyTorch>=1.3.0

* torchvision>=0.5.0

* scikit-learn>=0.19.2

* tqdm>=4.48.2

* learn2learn on the master branch

### Running ATG

A standalone re-implementation of ATG is provided in `examples/atg.py`. To run it on a synthetic dataset:

```bash

python examples/atg.py

```

### Training on ATG Partitions

~~~bash

python examples/train.py --algorithm='protonet' --dataset='mini-imagenet' --taskset='original'

~~~

where

* `taskset` takes values `easy`, `medium-easy`, `medium-hard`, `hard` or `randomX` where `X` is the seed to reproduce random splits.

* `dataset` takes values `mini-imagenet`, `tiered-imagenet`, `emnist`, `lfw10`, `cifar100`.

* `algorithm` takes values `protonet`, `maml`, `anil`.

For more help on the interface, run: `python examples/train.py --help`.