strisunshinewentingwang/pytorch

Pull Requests Commits

[vulkan] vTensor class cleanup (#80718)

SS-JIA committed 2 years ago

b4cdefdb

[vulkan] Remove old Resource Struct (#80717)

SS-JIA committed 2 years ago

50bbd9c0

[vulkan] Remove references to old Resource struct (#80716)

SS-JIA committed 2 years ago

5a0b4717

[vulkan] Replace `resource().pool.uniform(block).object` with `UniformParamsBuffer` (#80708)

SS-JIA committed 2 years ago

a49dfd83

Add prims::{real, imag}, refs::{real, imag}, remove prims::is_infinite (#80148)

zasdfgbnm committed 2 years ago

9ca561cb

[vulkan] Integrate refactored memory mapping, remove Tensor Future pattern (#80707)

SS-JIA committed 2 years ago

3f5b33a4

[vulkan] Integrate refactored fence (#80706)

SS-JIA committed 2 years ago

b6076471

[vulkan] Integrate refactored buffer and image (#80705)

SS-JIA committed 2 years ago

b52d51c5

[vulkan] Add refactored Resource (#80704)

SS-JIA committed 2 years ago

47fc108b

[vulkan] Remove ThreadContext (#80703)

SS-JIA committed 2 years ago

d275c3f4

[vulkan] Move ownership of Shader and Pipeline caches to Adapter (#80702)

SS-JIA committed 2 years ago

d80a31ed

[vulkan] Refactor Adapter to have PhysicalDevice hold physical device properties (#80701)

SS-JIA committed 2 years ago

e6672c2e

[vulkan] Refactor Shader and Pipeline (#80700)

SS-JIA committed 2 years ago

d6def7a8

[vulkan] Remove unused code for Winograd convolutions (#80699)

SS-JIA committed 2 years ago

3a295e92

Consolidate the naming of named_parameter and state_dict for CheckpointWrapper (#80089)

fegin committed 2 years ago

e0eeb06e

[Profiler] Add heuristic to rank events base on computed metrics (#80094)

davidchencsl committed 2 years ago

00710089

[Prims] Unbreak CUDA lazy init (#80899)

malfet committed 2 years ago

b62209f0

[ROCm] Load ROCm if Torch is used as a dependency (#80469)

BeyondPerception committed 2 years ago

d03f989d

Handling the error codes of hipGetDeviceCount (#80405)

pruthvistony committed 2 years ago

6bf4c662

[quant] Refactor quantize clamping into float_to_apot util function (#80885)

asl3 committed 2 years ago

5b493ba1

Fix the derivative of `acosh` for complex numbers. (#80841)

lezcano committed 2 years ago

be107879

[lint] create a workflow consistency linter (#80200)

suo committed 2 years ago

769df743

[test stats] record invoking file in test cases (#80792)

suo committed 2 years ago

769b446f

Add cached nvFuser's fusion creation for torch._prims.executor (#80525)

IvanYashchuk committed 2 years ago

9a12aa6c

Only run test on CUDA

SherlockNoMad committed 2 years ago

7e746bfd

[Expanded Weights] Fix instance norm (#79800)

samdow committed 2 years ago

56f4b69c

Fixed docstring typo (#80879)

vfdev-5 committed 2 years ago

8fc8713f

[Quant] Add fast path of qmean/qstd for quantized CPU (reopen #70172) (#80579)

Xia-Weiwen committed 2 years ago

54217e69

Fix `Module.share_memory` error (#80843)

kurtamohler committed 2 years ago

4c279994

fix docstring of PostLocalSGDOptimizer (#80855)

wayi1 committed 2 years ago

f76bb882

Newer Older