snap-stanford / comet Goto Github PK

View Code? Open in Web Editor NEW

111.0 9.0 14.0 1.31 MB

[ICLR 2021] Concept Learners for Few-Shot Learning

Home Page: https://snap.stanford.edu/comet/

License: Other

Python 99.86% Shell 0.14%

comet's Introduction

Concept Learners for Few-Shot Learning

Kaidi Cao*, Maria Brbić*, Jure Leskovec

Project website

This repo contains the reference source code in PyTorch of the COMET algorithm. COMET is a meta-learning method that learns generalizable representations along human-understandable concept dimensions. For more details please check our paper Concept Learners for Few-Shot Learning (ICLR '21).

Dependencies

The code is built with following libraries:

Getting started

CUB dataset

Change directory to ./filelists/CUB
Run source ./download_CUB.sh

Tabula Muris dataset

Change directory to ./filelists/tabula_muris
Run source ./download_TM.sh

Usage

Training

We provide an example here:

Run python ./train.py --dataset CUB --model Conv4NP --method comet --train_aug

Testing

We provide an example here:

Run python ./test.py --dataset CUB --model Conv4NP --method comet --train_aug

Tabula Muris benchmark

If you would like to test your algorithm on the new benchmark dataset introduced in our work, you can download the data as described above or directly at http://snap.stanford.edu/comet/data/tabula-muris-comet.zip.

Dataset needs to be preprocessed using preprocess.py. Train/test/validation splits are available in load_tabula_muris.

Running this code requires anndata and scanpy libraries.

Citing

If you find our code useful, please consider citing:

@inproceedings{
    cao2021concept,
    title={Concept Learners for Few-Shot Learning},
    author={Cao, Kaidi and Brbi\'c, Maria and Leskovec, Jure},
    booktitle={International Conference on Learning Representations (ICLR)},
    year={2021},
}

Our codebase is developed based on the benchmark implementation from paper A Closer Look at Few-shot Classification.

Tabula Muris benchmark is developed based on the mouse aging cell atlas from paper https://www.nature.com/articles/s41586-020-2496-1.

comet's People

Contributors

Stargazers

Watchers

Forkers

rvinas arita37 ruizhaoz e96031413 amadeusine youmna-salah tor4z gstoica27 gary828 me714 aesdee julianrein ashah03 avivnur

comet's Issues

error with training example on CUB

Running :
python ./train.py --dataset CUB --model Conv4NP --method comet --train_aug

Error :

File "/content/drive/MyDrive/FYP/comet-master/comet-master/CUB/data/dataset.py", line 68, in init
with open(data_file, 'r') as f:
FileNotFoundError: [Errno 2] No such file or directory: './filelists/CUB/base.json'

数据集的获取

求Reuters数据集和 Flowers数据集（概念处理的文件或者处理好的数据集）

Question on training with TM dataset

I used the following command to train the model with TM dataset
python ./train.py --model Conv4NP --method comet --train_aug

But, the following errors occur

Traceback (most recent call last):
  File "./train.py", line 13, in <module>
    from data.datamgr import SimpleDataManager, SetDataManager
  File "/root/notebooks/nfs/work/yanwei.liu/comet/TM/data/__init__.py", line 1, in <module>
    from . import datamgr
  File "/root/notebooks/nfs/work/yanwei.liu/comet/TM/data/datamgr.py", line 7, in <module>
    import data.additional_transforms as add_transforms
ModuleNotFoundError: No module named 'data.additional_transforms'

Thus I read comet/TM/data__init__.py. Line 3 from . import additional_transforms so I checkcomet/TM/data/ folder, there's no any .py file named additional_transforms. Did you forget to push the file to GitHub? or the command I used wasn't correct?

Thanks

PS. I have used the process.py to preprocess the dataset according to your guideline in README file.

Issues downloading CUB_200 dataset

Congratulations on the nice work!

I'm having some issues downloading the CUB 200 dataset with the download_CUB.sh script. It seems that the wget command does not work for Google Drive folders. Instead, the following command seems to work:

gdown https://drive.google.com/uc?id=1hbzc_P1FuxMkcabkgn9ZKinBwW683j45

Pre-trained model release

Hello,

are there any plans to release a model checkpoint used in the paper? Would be much appreciated.

Thanks.