mmaction
mmaction copied to clipboard
bad distributed
@zhaoyue-zephyrus @loveunk @lindahua @acrophile @yjxiong When I test the I3D model, I found that the name of the data read by each GPU is the same. Have you encountered the same situation and would you tell me how to solve it? Is this the reason why the program mentioned at https://github.com/open-mmlab/mmaction/issues/141 is stuck?Thanks.
Hi, what do you mean by 'the name of the data read by each GPU', do you mean 'img_group_0'?
@kennmckormick I means that the picture read by gpu 0 is the same as the picture read by gpu1.