search

uncbiag / SimpleClick

SimpleClick: Interactive Image Segmentation with Simple Vision Transformers (ICCV 2023)
MIT License
209 stars 32 forks source link

The batchsize for training SimpleClick-xTiny #33

Closed ChenyiZhang007 closed 4 months ago

ChenyiZhang007 commented 4 months ago

Hi, i didn't find this setting in the paper.

qinliuliuqin commented 4 months ago

https://github.com/uncbiag/SimpleClick/blob/f33112e5a37ac97b9adfb9cda0106830e77b5b7d/models/iter_mask/plainvit_xtiny448_cocolvis_itermask.py#L61

For xTiny, I used the default batch size, so the training has more iterations than ViT-Base.

ChenyiZhang007 commented 4 months ago

thanks a lot