search

jacksonloper / bardensr

barcode demixing through nonnegative spatial regression
MIT License
6 stars 5 forks source link

InvalidArgumentError: cannot compute Sub as input #1(zero-based) was expected to be a float tensor but is a double tensor #24

Closed tsuijenk closed 6 months ago

tsuijenk commented 6 months ago

image

jacksonloper commented 6 months ago

Looks like you might be feeding it float64 instead of float32.

Make sure all arrays are float32 and try again?

tsuijenk commented 6 months ago

Dear Jackson,

Thank you for your quick reply! Xnorm was already float32 at the time the issue was created. Are there any dtypes required for codeflat?

On Mon, Mar 18, 2024 at 5:06 PM Jackson Loper @.***> wrote:

Looks like you might be feeding it float64 instead of float32.

Make sure all arrays are float32 and try again?

— Reply to this email directly, view it on GitHub https://github.com/jacksonloper/bardensr/issues/24#issuecomment-2005382770, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKWVXJLSXXOLLG4PVZDRYB3YY56RXAVCNFSM6AAAAABE4RAIDKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMBVGM4DENZXGA . You are receiving this because you authored the thread.Message ID: @.***>

-- Very respectfully,

Jenkin Tsui, M.A., GStat. | Ph.D. Candidate

Department of Molecular Oncology L.J. Blackmore Cancer Research Centre @. | @.