strisunshinewentingwang/pytorch

Pull Requests Commits

[fix] max_pool1d: shape check (#85594)

kshitij12345 committed 1 year ago

8fb470e8

catches failure on nvprim speculative lowering (#85580)

jjsjann123 committed 2 years ago

cab6ffa0

Stop cuda-10.2 binary builds (#85873)

atalman committed 2 years ago

a807f198

Add opt-einsum to CI (#85574)

janeyx99 committed 2 years ago

3cdf621f

Release GIL when doing shared memory copies on Tensors (#85389)

ezhang887 committed 2 years ago

b25a1ce2

Revert "C10D extension to enable per-thread PG (#84153)"

pytorchmergebot committed 2 years ago

6fae62b3

Separate magma installation for ROCm into its own file (#85567)

jithunnair-amd committed 2 years ago

976e2a35

Treat layout / pin_memory consistently across creation refs (#85333)

lezcano committed 2 years ago

9fb72ca4

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

peterbell10 committed 2 years ago

a76995e5

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

Peizhao Zhang committed 2 years ago

ad87365e

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

slgong-fb committed 2 years ago

3cfc61b8

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

wanchaol committed 2 years ago

72b32f16

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

justinchuby committed 2 years ago

dc63948d

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

fufeisi committed 2 years ago

3c9e8cd8

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

slgong-fb committed 2 years ago

7628603a

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

BowenBao committed 2 years ago

edb99df2

Improve codegen for jvp decomposition (#84894)

soulitzer committed 2 years ago

7e468400

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

Taylor Robie committed 2 years ago

f23f362c

[Profiler] Fix traversal utility (#85717)

Taylor Robie committed 2 years ago

282d8dfa

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

Taylor Robie committed 2 years ago

dfdfaec3

Properly fix log_sigmoid vmapjvp and remove hack (#84892)

soulitzer committed 2 years ago

bd65adf4

Add bfloat16 support for lerp on CPU (#84327)

CaoE committed 2 years ago

cca90964

[ONNX] Update `unconvertible_ops` (#85595)

justinchuby committed 2 years ago

7cdd39b3

Implement duck shaping on SymInts (#85808)

ezyang committed 2 years ago

ada6e5b5

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

Xia-Weiwen committed 2 years ago

3a3e2002

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

zaf committed 2 years ago

d542aab5

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

eellison committed 2 years ago

6a2b12dd

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

zou3519 committed 2 years ago

a67621a6

Excise functorch/version.txt (#85830)

zou3519 committed 2 years ago

49859146

Disallow saved tensor hooks in functorch transforms (#85829)

zou3519 committed 2 years ago

d8277d90

Newer Older