Validate user-supplied outputSizes
of MLConvTranspose2dOptions
and MLPool2dOptions
#818
Labels
outputSizes
of MLConvTranspose2dOptions
and MLPool2dOptions
#818
This issue is tracking discussion in PR review #814 (comment) and #814 (comment).
The current algorithm steps of
convTranspose2d
and create pooling operation don't validateoutputSizes
attribute when it exists inMLConvTranspose2dOptions
andMLPool2dOptions
.As a reference, Chromium prototype validates
MLConvTranspose2dOptions.outputSizes
andMLPool2dOptions.outputSizes
./cc @inexorabletash
The text was updated successfully, but these errors were encountered: