strisunshinewentingwang/pytorch

Pull Requests Commits

[BE] use DeviceIndex instead of int64_t for related device interfaces (#103068)

cyyever committed 1 year ago

d9fb7166

Fixes #107737 SGD doc blank line (#107738)

janeyx99 committed 1 year ago

4656e094

Revert "Remove remaining global `set_default_dtype` calls from tests (#107246)"

pytorchmergebot committed 1 year ago

161ea463

[ATen] Update pre-compiled header (#106915)

peterbell10 committed 1 year ago

c68d0a70

Remove some unnecessary <iostream> includes from headers (#106914)

peterbell10 committed 1 year ago

a6c29b72

[activation checkpointing] Add default autocast keys to functional rng wrappers (#107934)

anijain2305 committed 1 year ago

78a053ba

Add backward check for test_memory_format (#106104)

CaoE committed 1 year ago

3992450e

[Quant][Inductor] Enable quantization conv_unary(relu) pattern fusion inside inductor (#105455)

leslie-fang-intel committed 1 year ago

c1e0fb7f

[Quant][Inductor] Enable dequant promotion inside inductor (#104590)

leslie-fang-intel committed 1 year ago

4f3ff16b

Implement size checking for copy_ with meta tensors (#107779)

Anstow committed 1 year ago

087c0613

[Quant][Inductor] Enable quantization conv pattern fusion inside inductor (#104588)

leslie-fang-intel committed 1 year ago

46f63e28

Fix how DDPOptimizer clones dynamo callback (#107834)

wconstab committed 1 year ago

572bc481

[Quant][Inductor] Enable quantized conv weight prepack inside inductor constant folding (#104581)

leslie-fang-intel committed 1 year ago

25678e31

Support cond and out_dtype for predispatch (#107941)

tugsbayasgalan committed 1 year ago

2b7271c7

[Quant][PT2E] Enable qconv for quantization 2.0 export (#104580)

leslie-fang-intel committed 1 year ago

8ef05725

[cuda] Fix the incorrect types in int8_gemm (#107895)

xwang233 committed 1 year ago

679e8e9d

[inductor] Add cat + split_with_sizes elimination pass (#107956)

aakhundov committed 1 year ago

d24c457b

Hide `transform` method by renaming it (#107940)

gmagogsfm committed 1 year ago

9af0e476

Added normal op decomposition for specializations of the normal op (#106792)

Anstow committed 1 year ago

598babf0

Revert "[Dynamo] cache_size policy (#107496)"

pytorchmergebot committed 1 year ago

b4c6c4da

[export] Store the arguments used to trace the exported program in itself (#107906)

angelayi committed 1 year ago

4b44b186

[ROCm] enable missed cpp tests - test_libtorch_jit (test_jit and test_lazy) (#107234)

hongxiayang committed 1 year ago

48e05d5d

[core aten] Remove split.Tensor from core aten (#107938)

angelayi committed 1 year ago

b382d553

Cleanup RUNNER_TEMP folder (#107868)

DanilBaibak committed 1 year ago

b445ed31

Revert "[optim] Make casting to match params a hook (#106725)"

pytorchmergebot committed 1 year ago

3a3cf0e0

Initial Python 3.12 build fixes (#106083)

albanD committed 1 year ago

b9472dec

[ONEDNN][BC-breaking] update onednn from v2.7.3 to v3.1.1 (#97957)

Xia-Weiwen committed 1 year ago

97a291f6

Enable custom device support in fsdp checkpoint (#107289)

dilililiwhy committed 1 year ago

ff37f601

Bump scipy from 1.9.3 to 1.10.1 in /.ci/docker (#104746)

dependabot[bot] committed 1 year ago

Verified b18e1b68

[ROCm] Use rocm manylinux builder image for triton wheels (#107600)

jataylo committed 1 year ago

196ef78b

Newer Older