strisunshinewentingwang/pytorch

Pull Requests Commits

doc(typo): memroy -> memory (#89126)

rmax committed 1 year ago

3af5cf4d

Use the Python frame safely in _pythonCallstack (#88993)

charlie-wt committed 1 year ago

cfd55254

handle scatter(Scalar) overload in inductor (#88894)

Krovatkin committed 1 year ago

8506b305

Fix nightly build binary errors (#89153)

atalman committed 1 year ago

0c835e25

[ONNX] Add onnx-script test cases (#86907)

titaiwangms committed 1 year ago

98379a39

Use torchrun for dynamo/distributed.py (#89149)

wconstab committed 1 year ago

f920bfaf

add test_c10d_spawn_ucc.py (#86508)

Fuzzkatt committed 1 year ago

8ba62bdf

Fix inaccuracy in nt constructor documentation + broken rendering (#89152)

mikaylagawarecki committed 1 year ago

ec61951f

Removed unecessary check in `select_nested` (#89150)

mikaylagawarecki committed 1 year ago

5848704e

[FSDP] Add fast path for `NO_SHARD` `clip_grad_norm_()` (#89137)

awgu committed 1 year ago

ee1d375b

[Dynamo] Fix bug in NamedTupleVariable (#89110)

yanboliang committed 1 year ago

e70f446a

More dynamo dashboard improvements (#89155)

williamwen42 committed 1 year ago

640af8d7

Fix XLASymNode.str() no str() attribute error (#89093)

Krovatkin committed 1 year ago

305b9b1f

Reland "SymIntify convolution backend calculation (#89069)"" (#89142)

ezyang committed 1 year ago

4908a125

[ao] making _is_activation_post_process private (#87520)

HDCharles committed 1 year ago

45c62a33

[PT-D][Checkpointing] Move distributed checkpointing from torch.distributed._shard.checkpoint to torch.distributed.checkpoint (#88698)

wz337 committed 1 year ago

aee96bbf

Prevent module full_backward_hook from erroring in double backward (#88357)

soulitzer committed 1 year ago

6b521bbf

[ONNX] Document ONNX diagnostics (#88371)

BowenBao committed 1 year ago

05813319

[Dynamo] Add a dummy profiler to avoid activating real profiler (#88930)

yanboliang committed 1 year ago

848e7240

[Quant][bc-breaking] Remove overwrite_output_observer (#88620)

andrewor14 committed 1 year ago

61801799

Support test-config filter logic for rocm (#89046)

huydhn committed 1 year ago

a6ef2c76

Run tests from test/inductor in inductor CI job (#88957)

peterbell10 committed 1 year ago

7b0adc29

Add bfloat16 support to torch.prod to align with torch.cumprod (#87205)

lezcano committed 1 year ago

58ebf92c

Fix decomp for embedding_backward and simplify the decomposition of embedding_dense and embedding_dense_backward (#87204)

lezcano committed 1 year ago

33209153

Simplify linspace decomp and increase its tolerance (#87203)

lezcano committed 1 year ago

e1ecf53d

test_unary_ufuncs few tests enabled on rocm which are passing (#89007)

bmedishe committed 1 year ago

d2d22d89

add quantize_decomposed_dynamic to op lib (#88855)

JacobSzwejbka committed 1 year ago

7f55db4f

Revert "Towards unifying symbolic and non symbolic fake tensor (#89038)"

pytorchmergebot committed 1 year ago

cf6003f0

[UCC] Add pre & post processing for CPU collectives (#89030)

kirteshpatil committed 1 year ago

fe276ea0

Revert "SymIntify convolution backend calculation (#89069)"

pytorchmergebot committed 1 year ago

90db86be

Newer Older