strisunshinewentingwang/pytorch

Pull Requests Commits

[vulkan] Fix concat op in feature dimension (#101721)

SS-JIA committed 1 year ago

c6180936

Have irange use `if constexpr` (#94050)

r-barnes committed 1 year ago

be94ff97

bring quantized_backward.cpp in sync with intern (#101990)

izaitsevfb committed 1 year ago

38e73b30

[torch.library] Change Library.__del__ into weakref.finalize (#101829)

zou3519 committed 1 year ago

4de5ee43

Add documentation for a catching invalid index type (#96451)

pavanky committed 1 year ago

5e635e17

Fix device normalization of automatically generate methods for custom backends. (#101796)

fakeYan committed 1 year ago

c9f8f4cf

[Reland] fix missing-prototypes warnings in torch_cpu (Part 5) (#101976)

cyyever committed 1 year ago

4db2dade

[c10d] Fix the check message of unsupported collectives ops. (#101775)

shaoyf42 committed 1 year ago

bdb3fb49

Suppress profiler spam in dynamo benchmarks (#101942)

jansel committed 1 year ago

5ba16011

[dtensor][2/N] more tensor ops to use strategy propagation (#101203)

wanchaol committed 1 year ago

38a29324

Revert "group constraints by arg (#101815)"

malfet committed 1 year ago

496212f4

Fix Backend docs search items (#101214)

ain-soph committed 1 year ago

a6303286

Hint Tensor._make_subclass as a staticmethod (#101961)

ringohoffman committed 1 year ago

a6f4088c

Explain how fastAtomicAdd works (#101951)

ezyang committed 1 year ago

19af5c0b

[Reland] fix missing-prototypes warnings in torch_cpu (Part 4) (#101949)

cyyever committed 1 year ago

4f2c007a

Revert "[dynamo] Minor refactor to use is_allowed to decide inlining of NNModule methods (#101910)"

pytorchmergebot committed 1 year ago

d0bb8fdc

Update Kineto submodule (#101952)

openrichardfb committed 1 year ago

e9a71156

[inductor] fix avg_pool2d accuracy problem in lowering (#101789)

Valentine233 committed 1 year ago

0a694dba

torch.unique with dim: NumPy compatible sorting (#101693)

nikitaved committed 1 year ago

3004d404

show errors on bazel test failure (#101928)

dagitses committed 1 year ago

dcffd5c6

improve diagnostics from bazel_linter.py (#101445)

dagitses committed 1 year ago

2a62b59e

use bazelisk as the bazel binary for lintrunner (#101744)

dagitses committed 1 year ago

b54cdaf9

[CUDA][CUBLAS] Fix BF16 reduced precision reduction note in Numerical accuracy docs (#101884)

lkct committed 1 year ago

807d8115

[export] Prototype on serialization schema. (#101899)

zhxchen17 committed 1 year ago

351c2ea2

[MPS] Fix embedding cache key (#101857)

qqaatw committed 1 year ago

330c9073

Partially fix shape mismatch in vision_maskrcnn (#101477)

ezyang committed 1 year ago

22ca1a11

[vision hash update] update the pinned vision hash (#101938)

pytorchupdatebot committed 1 year ago

9e8da7fb

[T153220354] Fix header inclusions in c10 (#1541) (#101846)

q10 committed 1 year ago

66a2600b

Prevent pattern matches across mutation ops in inductor pre-grad FX passes (#101144)

williamwen42 committed 1 year ago

dde6d561

disableing quantizing gradient in 8bw (#101739)

kwanghoon-meta committed 1 year ago

13640bf9

Newer Older