-
Notifications
You must be signed in to change notification settings - Fork 208
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Reshaping CuArray throws error during backpropagation #162
Comments
The core issue in an ambiguity with a GPUArrays method – |
Where is the CuArray |
https://github.com/JuliaGPU/CuArrays.jl/blob/bd5f82ca1ca7a2bd8f20d6087308ba4cdb44c6e7/src/array.jl#L126-L131 is where the ambiguous definition used to be. |
Since it's just the identity, the same array is just returned. Handles https://github.com/JuliaGPU/CuArrays.jl/issues/161 , SciML/DiffEqFlux.jl#21 .
282: Fix ambiguity in _reshape when dims is Tuple{Int} r=maleadt a=ChrisRackauckas Since it's just the identity, the same array is just returned. Handles https://github.com/JuliaGPU/CuArrays.jl/issues/161 , SciML/DiffEqFlux.jl#21 . @vchuravy Co-authored-by: Christopher Rackauckas <accounts@chrisrackauckas.com>
This in principle should be fixed. At least, tests now show that reshaping a CuVector to itself now works without ambiguity. We can either close or directly test this on Flux. |
System information
The forward propagation contains reshape operations like below
These operations work fine on CPU during forward and back propagation.
On GPU it causes the following error
The text was updated successfully, but these errors were encountered: