search

Epiphqny / SOLOv2

SOLOv2: Dynamic, Faster and Stronger, achives 39.5mAP on coco test-dev (36 epochs result)
https://arxiv.org/abs/2003.10152
Apache License 2.0
245 stars 37 forks source link

training batch size #11

Closed twangnh closed 4 years ago

twangnh commented 4 years ago

Hi, here the config uses batch size 4 for training, so in total 32 for 8-gpu machine, but batch size in solov2 paper is 16, I'm wondering if bs=16 has lower performance, or similar with bs=32 and you just want to training faster? Thanks !

Epiphqny commented 4 years ago

Just want to training faster.

twangnh commented 4 years ago

may be bs=16 could have better performance since bs is smaller?

Epiphqny commented 4 years ago

You can try and test.

twangnh commented 4 years ago

thanks