strisunshinewentingwang/pytorch

Pull Requests Commits

triu/tril: complete dtype support for CPU/CUDA. (#101414)

nikitaved committed 1 year ago

c51dfbf5

inductor: fix sign gets wrong result dtype issue (#101377)

XiaobingSuper committed 1 year ago

88b6a457

[profiler] When record_inputs=True, record scalar lists of length <= 30 (#100593)

davidberard98 committed 1 year ago

935100cb

inductor: add dtype check before doing cpu binary fusion (#101376)

XiaobingSuper committed 1 year ago

e389bfa0

[inductor] enable descriptive name for cpp kernels (#101330)

jgong5 committed 1 year ago

6f7ebcdc

[inductor] move dtype propagation log to schedule artifact (#101351)

jgong5 committed 1 year ago

86869475

[inductor] add graph id prefix to inductor_wrapper_call profile info (#101350)

jgong5 committed 1 year ago

dfc46153

[vision hash update] update the pinned vision hash (#101471)

pytorchmergebot committed 1 year ago

d9d34b3e

add retries to external contribution data upload (#100889)

PaliC committed 1 year ago

c03555a3

[ONNX] Diagnostic to show all unsupported call_functions (#100451)

titaiwangms committed 1 year ago

773f6b62

[ONNX] Diagnostic 'log' and 'log_and_raise_if_error' (#100407)

titaiwangms committed 1 year ago

45d080e0

Re-enable detectron2_maskrcnn on CI (#100791)

yanboliang committed 1 year ago

e4eaf333

Update the torchbench pin to include timm upgrade (#101466)

xuzhao9 committed 1 year ago

af4248b9

[quant][pt2] Handle no conv bias in prepare QAT fusion (#100610)

andrewor14 committed 1 year ago

964e61ee

[Dynamo] Improve handling UnspecializedNNModuleVariable side effect (#101141)

yanboliang committed 1 year ago

7052fb37

[c10d][Fix] Start gloo sequence numbers at 0. (#101422)

rohan-varma committed 1 year ago

6f8a71aa

[IValue] Only coalesce once (#101447)

rohan-varma committed 1 year ago

4b849744

Revert "[fx] Better replacements finder in subgraph rewriter (#100556)"

pytorchmergebot committed 1 year ago

13056ca2

Add more canonical way of adding runtime pass (#100956)

tugsbayasgalan committed 1 year ago

194d3603

Use %zu instead of %ld when formatting size_t (#101412)

pandaninjas committed 1 year ago

f0786ad7

Clean up grad check in sdp_utils.h (#101435)

drisspg committed 1 year ago

52363de2

Use python3 instead of /usr/bin/env python3 on Windows (#101437)

huydhn committed 1 year ago

c3f7db3f

Fix Wishart distribution documentation (#95816)

alexisthual committed 1 year ago

24cc7fe0

[Inductor] Improve view/reshape on tensors with shape 0 (#101051)

yanboliang committed 1 year ago

7f3b00bf

Revert torch.fx.interpreter error printing change (#101462)

ezyang committed 1 year ago

d1980336

[caffe2/tools/autograd] Fix non-determinism in code gen (#101425)

andrewjcg committed 1 year ago

799ef7e5

fix print tensor in cpp for privateuse1 (#100797)

heidongxianhua committed 1 year ago

a8376099

[decomp] fix decomp of batch_norm when weight/bias is not flattened (#101059)

jgong5 committed 1 year ago

788ff062

Inductor cpp wrapper: support sympy.Expr as input (#101257)

chunyuan-w committed 1 year ago

1faef895

Enable default workflow PyT 2.0 UTs on ROCm stack (#100981)

jataylo committed 1 year ago

187eb7ca

Newer Older