Jiazhen Li

Results 3 issues of Jiazhen Li

How can I resolve this? I used cuda 11.8+ torch 2.0.1 + python 3.9 and this problem still exists.(my environment)

I'm using the default settings, but due to GPU memory issues, batch_size=4, in later evaluation phases, the aps of the ten classes are small, wondering if the reason is due...