search

yyliu01 / PS-MT

[CVPR'22] Perturbed and Strict Mean Teachers for Semi-supervised Semantic Segmentation
https://arxiv.org/pdf/2111.12903.pdf
MIT License
187 stars 17 forks source link

About training config #9

Closed Yan1026 closed 2 years ago

Yan1026 commented 2 years ago

Hi,thank you for your great work! If I use 4GPU,and batchsize=16 learning-rate=1e-2 ,will it be the best configuration?

And why you said 2GPU can get better performance rather than 4GPU?

yyliu01 commented 2 years ago

Hi,

I've only tested the 1464 high-quality setting with 2GPUs, and it outperforms the paper results by about 1 point. Please follow the settings in here (https://wandb.ai/pyedog1976/PS-MT(VOC12)/runs/23b8qul0/overview?workspace=user-pyedog1976).

I haven't tested it for the aug setting, so it will be safe to run the setting provided by the scripts.

Regards, Yuyuan

Yan1026 commented 2 years ago

I run with ./scripts/train_voc_aug.sh -l 1323 -g 4 -b 101 but get error：

ID 3 Warm (4) | Ls 0.51 |: 98%|█████████████████████████████████████████████████████████████████████▎ | 40/41 [01:04<00:00, 1.18it/s] ID 3 Warm (4) | Ls 0.51 |: 98%|█████████████████████████████████████████████████████████████████████▎ | 40/41 [01:14<00:00, 1.18it/s] ID 3 Warm (4) | Ls 0.51 |: 100%|███████████████████████████████████████████████████████████████████████| 41/41 [01:14<00:00, 3.65s/it] ID 3 Warm (4) | Ls 0.51 |: 100%|███████████████████████████████████████████████████████████████████████| 41/41 [01:14<00:00, 1.81s/it]

0%| | 0/289 [00:00<?, ?it/s]wandb: Network error (ConnectionError), entering retry loop.

How can I solve this problem？

yyliu01 commented 2 years ago

It doesn't need to be fixed. Wandb sometimes encounters network traffic issues, but it will be reconnected automatically and will not affect the training.