Pytorch implementation of UniNet (ECCV 2022)
This repo is the offcial implementation of the paper UniNet: Unified Architecture Search with Convolution, Transformer, and MLP
@article{UniNet,
author = {Jihao Liu, Xin Huang, Guanglu Song, Yu Liu, Hongsheng Li},
journal = {arXiv:2207.05420},
title = {UniNet: Unified Architecture Search with Convolution, Transformer, and MLP},
year = {2022},
}
20/12/2022 Update pretrained models.
25/10/2022 Update the source code.
The code is tested with torch==1.11
and timm==0.5.4
.
Models | Params (M) | FLOPs (G) | Pretrain Epochs | Top-1 Acc. | ckpt |
---|---|---|---|---|---|
UniNet-B1 | 11.5 | 1.1 | 300 | 81.0 | ckpt |
Currently, we supporting running experiments with slurm. You can reproduce the results of UniNet-B1 as follows:
sh exp/b1/run.sh partition 8