Awesome

LRA-IGLOO

Expected results

Those are the results one is expected to find using IGLOO on the Long Range Arena suite of benchmarks.

@inproceedings{
tay2020long,
title={Long Range Arena : A Benchmark for Efficient Transformers },
author={Yi Tay, Mostafa Dehghani, Samira Abnar, Yikang Shen, Dara Bahri,
Philip Pham, Jinfeng Rao, Liu Yang, Sebastian Ruder, Donald Metzler},
booktitle={ArXiv Preprint},
year={2020},
url={},
note={under review}
}

Prerequisites

Tensorflow 2.0.0
Tensorflow-datasets 2.0.0
Numpy

Building

To install the prerequisites.

$ pip install -r requirements.txt

Running

$ python longrangearena_cifar10.py

$ python longrangearena_imdb.py

$ python longrangearena_matching.py

$ python pathfinder/longrangearena_pathfinder.py

$ python lra_listops/longrangearena_listops.py