pytorch

f85f58bf - Fix quantized linear vulkan tests (#120960)

Fix quantized linear vulkan tests (#120960) Summary: Fixed quatized linear vulkan tests by using an old pack_biases function. Test Plan: **Vulkan quantized api tests** buck2 run --target-platforms ovr_config//platform/macos:arm64-fbsource //xplat/caffe2:pt_vulkan_quantized_api_test_binAppleMac\#macosx-arm64 -c pt.vulkan_full_precision=1 ... ... ... [ RUN ] VulkanAPITest.linear_2d_flat [ OK ] VulkanAPITest.linear_2d_flat (5 ms) [ RUN ] VulkanAPITest.linear_2d_small [ OK ] VulkanAPITest.linear_2d_small (0 ms) [ RUN ] VulkanAPITest.linear_2d_large [ OK ] VulkanAPITest.linear_2d_large (4 ms) [ RUN ] VulkanAPITest.linear_3d_flat [ OK ] VulkanAPITest.linear_3d_flat (2 ms) [ RUN ] VulkanAPITest.linear_3d_small [ OK ] VulkanAPITest.linear_3d_small (1 ms) [ RUN ] VulkanAPITest.linear_3d_large [ OK ] VulkanAPITest.linear_3d_large (1 ms) [ RUN ] VulkanAPITest.linear_4d_flat [ OK ] VulkanAPITest.linear_4d_flat (1 ms) [ RUN ] VulkanAPITest.linear_4d_small [ OK ] VulkanAPITest.linear_4d_small (1 ms) [ RUN ] VulkanAPITest.linear_4d_large [ OK ] VulkanAPITest.linear_4d_large (2 ms) ... ... [----------] 85 tests from VulkanAPITest (1704 ms total) [----------] Global test environment tear-down [==========] 85 tests from 1 test suite ran. (1704 ms total) [ PASSED ] 85 tests. YOU HAVE 8 DISABLED TESTS **Vulkan api tests** buck2 run --target-platforms ovr_config//platform/macos:arm64-fbsource //xplat/caffe2:pt_vulkan_api_test_binAppleMac\#macosx-arm64 -c pt.vulkan_full_precision=1 [----------] Global test environment tear-down [==========] 426 tests from 1 test suite ran. (4997 ms total) [ PASSED ] 423 tests. [ SKIPPED ] 1 test, listed below: [ SKIPPED ] VulkanAPITest.querypool_flushed_shader_log [ FAILED ] 2 tests, listed below: [ FAILED ] VulkanAPITest.log_softmax_underflow [ FAILED ] VulkanAPITest.log_softmax Differential Revision: D54396367 Pull Request resolved: https://github.com/pytorch/pytorch/pull/120960 Approved by: https://github.com/yipjustin