strisunshinewentingwang/pytorch

Pull Requests Commits

Improve make_tensor performance for float and complex types (#85473)

peterbell10 committed 1 year ago

a76995e5

[qat]A more stable conv_bn fusion for qat training. (#85744)

Peizhao Zhang committed 1 year ago

ad87365e

[Profiler][trivial] Optimizer states (part 4 of Record Optimizer) (#85840)

slgong-fb committed 1 year ago

3cfc61b8

[c10d] move ncclgetlasterror directive definition upfront (#85825)

wanchaol committed 1 year ago

72b32f16

[ONNX] Update behavior for `register_custom_op_symbolic` (#85636)

justinchuby committed 1 year ago

dc63948d

Create a quantized non-in-palce version CUDA ReLU function, (#85669)

fufeisi committed 1 year ago

3c9e8cd8

[Profiler] bug fix: python object reference counting (#85847)

slgong-fb committed 1 year ago

7628603a

[ONNX] Fix reduce node shape inference (#85765)

BowenBao committed 1 year ago

edb99df2

Improve codegen for jvp decomposition (#84894)

soulitzer committed 1 year ago

7e468400

[Profiler] Use strong typedef for Tensor ID (#85718)

Taylor Robie committed 1 year ago

f23f362c

[Profiler] Fix traversal utility (#85717)

Taylor Robie committed 1 year ago

282d8dfa

[Profiler] Don't assign in AppendOnlyList::emplace_back (#85716)

Taylor Robie committed 1 year ago

dfdfaec3

Properly fix log_sigmoid vmapjvp and remove hack (#84892)

soulitzer committed 1 year ago

bd65adf4

Add bfloat16 support for lerp on CPU (#84327)

CaoE committed 1 year ago

cca90964

[ONNX] Update `unconvertible_ops` (#85595)

justinchuby committed 1 year ago

7cdd39b3

Implement duck shaping on SymInts (#85808)

ezyang committed 1 year ago

ada6e5b5

[Quant] Add unified x86 quant backend (#84329)

Xia-Weiwen committed 1 year ago

3a3e2002

[quant][ao_migration] nn.intrinsic migration to ao (#84842)

zaf committed 1 year ago

d542aab5

Turn on aliasing tests for fake backwards, Fix Batch norm running mean/var decomp aliasing (#85471)

eellison committed 1 year ago

6a2b12dd

Update functorch README to reflect move into PyTorch (#85832)

zou3519 committed 1 year ago

a67621a6

Excise functorch/version.txt (#85830)

zou3519 committed 1 year ago

49859146

Disallow saved tensor hooks in functorch transforms (#85829)

zou3519 committed 1 year ago

d8277d90

Add mechanism to disable the "saved tensors hooks" feature (#85553)

zou3519 committed 1 year ago

5aa183d2

[ONNX] Deprecate setter functions for global variables (#85165)

justinchuby committed 1 year ago

85d8441f

[ONNX] Annotate all g as GraphContext (#85491)

justinchuby committed 1 year ago

5deeb09d

[ONNX] Create decorator to handle symbolic context (#84776)

justinchuby committed 1 year ago

c42a408b

[cuDNN][cuDNN v8 API] Fix 3d convolution_add_relu in V8 (#85055)

eqy committed 1 year ago

723193ec

Allow only one -1 in nested view/reshape (#85691)

mikaylagawarecki committed 1 year ago

01add6e2

Fix upload condition pypi-cudnn build (#85799)

atalman committed 1 year ago

1418a663

[ONNX] Create GraphContext and load `g.op` method to the class (#84728)

justinchuby committed 1 year ago

3d231667

Newer Older