search

jonbarron / robust_loss_pytorch

A pytorch port of google-research/google-research/robust_loss/
Apache License 2.0
656 stars 88 forks source link

fixed shape mismatch in AdaptiveImageLossFunction #15

Open rvarun7777 opened 4 years ago

jonbarron commented 4 years ago

There are unit tests that need to be updated as well.

rvarun7777 commented 4 years ago

I could not find an issue in your unit tests.

jonbarron commented 4 years ago

I patched your PR and ran nosetests from the root directory, and it causes the testImageLossfun* tests to fail. Those tests need to also be rewritten with the same NHWC-> NCHW dimension reordering.

On Mon, Mar 9, 2020 at 3:40 AM Varun Ravikumar notifications@github.com wrote:

Tested your loss on YUV space on monodepth2 project. It just drives the final loss below 0.# AdaptiveImageLossFunction((3, 640, 192), torch.float32, "cuda:0", color_space='YUV', scale_lo=0.01, scale_init=0.01) These were the initial setting used. Final loss = 0.85 Adaptive Loss + 0.15 SSIM + 0.01 Smoothness

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jonbarron/robust_loss_pytorch/pull/15?email_source=notifications&email_token=AAGZFNUVPB7HDFIHYYCF5QLRGTBR3A5CNFSM4LD4YMAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEOGSF3Q#issuecomment-596452078, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAGZFNVMR43AEPPOXYNFGELRGTBR3ANCNFSM4LD4YMAA .