How do you compute the param of SRN-Deblur model?3.76M?

[Courageux-J]

[tejpratapgit]

Hello courageux,
How many params did you get. We observed ~6.8M params by excluding adam optimzer training params.

[jiangsutx]

There is a difference between released model and this table.
We mentioned in the paper that in this table conv kernels are 3x3, in order for fast experiments.
And in release model, kernels are 5x5.

[tejpratapgit]

Ok. Thank you. Regards, Tej

…

On Mon, 16 Dec 2019, 9:11 am Xin Tao, ***@***.***> wrote: There is a difference between released model and this table. We mentioned in the paper that in this table conv kernels are 3x3, in order for fast experiments. And in release model, kernels are 5x5. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#48?email_source=notifications&email_token=AFM3USTH2KK4HWNSY6O32NDQY32F5A5CNFSM4JZKLPU2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEG5NNQQ#issuecomment-565892802>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFM3USXE2HXEZ5UJVS3UBGLQY32F5ANCNFSM4JZKLPUQ> .

[noranali]

hello , could you help me to compute the model parameters please ?