Learning-Rate-Dropout
Pytorch implementation of Learning Rate Dropout.
Paper Link: https://arxiv.org/pdf/1912.00144.pdf
Train ResNet-34 for Cifar10:
run:
python main.py --model=resnet --optim=adam_lrd --lr=0.001 --LRD_p=0.5
python main.py --model=resnet --optim=adam --lr=0.001
python main.py --model=resnet --optim=sgd_lrd --lr=0.1 --LRD_p=0.5
python main.py --model=resnet --optim=sgd --lr=0.1
python main.py --model=resnet --optim=rmsprop_lrd --lr=0.001 --LRD_p=0.5
python main.py --model=resnet --optim=rmsprop --lr=0.001
After training, run "plot.py" to show the learning curves.
8 Sep 27, 2021
87 Dec 06, 2022
341 Dec 20, 2022
52 Dec 22, 2022
1 Jan 19, 2022
440 Jan 02, 2023
10 Jul 16, 2022
17 Apr 25, 2022
6 Jan 17, 2022
86 Dec 29, 2022
0 Feb 09, 2022
46 Dec 29, 2022
153 Nov 14, 2022
105 Dec 23, 2022
11.5k Jan 08, 2023
6.4k Dec 31, 2022
63 Dec 12, 2022
670 Dec 30, 2022
3 Dec 07, 2022
3.1k Jan 04, 2023