Open narsisn opened 4 weeks ago
Hi narsisn,
We implemented the segmentation task using the code from TinyMIM and the head is uper head.
We initialized the backbone with the pre-trained model from ImageNet. If you load the base_01346.pth, the associated pruning index is [0,1,3,4,6] too.
Best, Sihao
Hi, Thanks for sharing your great work. Which decoder have you used for semantic segmentation tasks? And could you please share the prune-layer for this task?
Thanks.