search

researchmm / TTSR

[CVPR'20] TTSR: Learning Texture Transformer Network for Image Super-Resolution
MIT License
765 stars 115 forks source link

question about SearchTransfer #56

Open SolitaryManF opened 1 year ago

SolitaryManF commented 1 year ago

SearchTransfer.py line 32: R_lv3 = torch.bmm(refsr_lv3_unfold, lrsr_lv3_unfold) #[N, HrWr, HW] torch.bmm takes up too much memory. RuntimeError: CUDA out of memory. Tried to allocate 1525.88 GiB Is there anything I haven't noticed?