@MzeroMiko Thank you for your outstanding work and generous open-source contributions! I have a question regarding the use of RandomSampler when drawing ERFs for each feature extraction backbone:
https://github.com/MzeroMiko/VMamba/blob/c07d2ea1322061cd52a505fdae735c91346a5d65/analyze/utils.py#L317
Doesn't this mean that each heatmap represents the average result of several different image batches? Or is this not a significant concern because each batch contains 50 images, which is sufficient to mitigate any bias? Thank you!
@MzeroMiko Thank you for your outstanding work and generous open-source contributions! I have a question regarding the use of
RandomSampler
when drawing ERFs for each feature extraction backbone: https://github.com/MzeroMiko/VMamba/blob/c07d2ea1322061cd52a505fdae735c91346a5d65/analyze/utils.py#L317 Doesn't this mean that each heatmap represents the average result of several different image batches? Or is this not a significant concern because each batch contains 50 images, which is sufficient to mitigate any bias? Thank you!