pytorch

b336deb6 - [quant][mobile] Not use qnnpack max_pool2d if ceil_mode is true (#34844)

[quant][mobile] Not use qnnpack max_pool2d if ceil_mode is true (#34844) Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/34844 QNNPACK max_pool2d operator does not support ceil_mode so this can cause crashes in the kernel when it is set to true. We default to the server implementation when ceil_mode is set to true Test Plan: python test/test_quantized.py Imported from OSS Differential Revision: D20478701 fbshipit-source-id: 7962444ac493f5c3c32a9aa1a7be465e8b84ccc2