ciflow/xpu/185606: Enable TF32 fpmath mode for XPU deconvolution
🔒
https://github.com
«The convolution path already calls apply_tf32_if_allowed() to set
oneDNN fpmath_mode::tf32 when the user enables it, but deconvolution
was missing this, causing ConvTranspose ops to always run in strict
fp32 mode regardl...»
Automatische Weiterleitung...
1.5s