CMSF
Official Code for "Constrained Mean Shift Using Distant Yet Related Neighbors for Representation Learning"
Requirements
- Python >= 3.7.6
- PyTorch >= 1.4
- torchvision >= 0.5.0
- faiss-gpu >= 1.6.1
Install PyTorch and ImageNet dataset following the official PyTorch ImageNet training code. We used Python 3.7 for our experiments.
- Install PyTorch (pytorch.org)
To run NN and CMSF-KM, you require to install FAISS.
FAISS:
- Install FAISS (https://github.com/facebookresearch/faiss/blob/master/INSTALL.md)
Training Self-Supservised CMSF-KM
python self_supervised/train_msf_km.py \
--cos \
--weak_strong \
--learning_rate 0.05 \
--epochs 200 \
--arch resnet50 \
--topk 5 \
--momentum 0.99 \
--mem_bank_size 128000 \
--num_clusters 50000 \
--checkpoint_path <CHECKPOINT PATH> \
<DATASET PATH>
Training Self-Supservised CMSF-2Q
python self_supervised/train_msf_2q.py \
--cos \
--weak_strong \
--learning_rate 0.05 \
--epochs 200 \
--arch resnet50 \
--topk 5 \
--momentum 0.99 \
--mem_bank_size 128000 \
--topkp 5 \
--checkpoint_path <CHECKPOINT PATH> \
<DATASET PATH>
Training Supservised
Following command can be used to train the CMSF(Supervised Learning)
python supervised/train_sup_msf.py \
--cos \
--weak_strong \
--learning_rate 0.05 \
--epochs 200 \
--arch resnet50 \
--topk 10 \
--momentum 0.99 \
--mem_bank_size 128000 \
--checkpoint_path <CHECKPOINT PATH> \
<DATASET PATH>
License
This project is under the MIT license.
25 Nov 15, 2022
2 Dec 27, 2021
3 Oct 31, 2022
1 Dec 23, 2021
1.2k Dec 27, 2022
103 Nov 29, 2022
137 Dec 20, 2022
7 Sep 23, 2022
7 Mar 22, 2022
36 Nov 04, 2022
2k Jan 01, 2023
7 Oct 27, 2022
4 Jan 07, 2023
173 Dec 26, 2022
5 Dec 18, 2022
36 Sep 08, 2022
40 Dec 27, 2022
186 Dec 20, 2022
8 Oct 08, 2022
6 Nov 29, 2022