-
Notifications
You must be signed in to change notification settings - Fork 8
Open
Description
Thank you for open-source this project!!!Now I can run the code for 1000 iters
[TRAIN] Iter: 200 Loss: 0.018154477700591087 PSNR: 20.443443298339844 TIME: 0h:0m:38.12s [TRAIN] Iter: 400 Loss: 0.0138796865940094 PSNR: 21.603029251098633 TIME: 0h:1m:16.01s [TRAIN] Iter: 600 Loss: 0.01028226688504219 PSNR: 22.845338821411133 TIME: 0h:1m:54.20s [TRAIN] Iter: 800 Loss: 0.008638457395136356 PSNR: 23.685611724853516 TIME: 0h:2m:32.25s [TRAIN] Iter: 1000 Loss: 0.006889075972139835 PSNR: 24.694589614868164 TIME: 0h:3m:10.62s
but get CUDA out of memory.
after that , which parameters should be changed to solve cut down the memory for the steps after 1000 ,maybe 1200
Metadata
Metadata
Assignees
Labels
No labels