strisunshinewentingwang/pytorch

Pull Requests Commits

clee2000 committed 2 years ago

2c4723af

clee2000 committed 2 years ago

49945a6b

clee2000 committed 2 years ago

fd2adf1e

[HIP] Add extra exception handling for non-ROCM builds (#83009)

Gamrix committed 2 years ago

30af17ce

[FSDP] Use _init_from_local_tensor to create ShardedTensor to avoid communication overhead (#82911)

fegin committed 2 years ago

24469020

Delayed compilation of backwards pass to when backwards runs (#83367)

Chillee committed 2 years ago

5e8b4c64

[quant] Optionally clamp weights post quantization (#83438)

digantdesai committed 2 years ago

1f7153be

expand torch.full to reason about integers (#83087)

migeed-z committed 2 years ago

ab02b898

fix bug in a linear constraint (#82938)

migeed-z committed 2 years ago

1a38724e

Revert "NestedTensor Softmax (#83435)"

pytorchmergebot committed 2 years ago

0061e676

made some minor tweaks to minifier to reduce outputs more often (#83565)

Chillee committed 2 years ago

eb4e03dd

Make sure that we can load old optimizer checkpoint (#83588)

albanD committed 2 years ago

84c4b079

Add docstring type formatting guidelines to `CONTRIBUTING.md` (#83536)

ProGamerGov committed 2 years ago

dcda9076

Add torch.ops.aten -> torch._refs mapping to TorchRefsMode using decomposition_table (#82657)

IvanYashchuk committed 2 years ago

9f03444f

[ROCm] Enable test_ddp_profiling_torch_profiler (#82749)

jaglinux committed 2 years ago

7af32084

Delete checked_dense_tensor_unwrap (#83543)

kit1980 committed 2 years ago

c8ec4ceb

Use opmath_type for CUDA logcumsumexp (#83425)

peterbell10 committed 2 years ago

822a8e05

[primTorch] support for a few magic methods (#83524)

Fabio Rocha committed 2 years ago

2a096e94

Make Adam optimizer differentiable (#82205)

Emilio Castillo committed 2 years ago

5aab57e1

[torchgen] Add logic in annotation parser to accept alias set (#83501)

larryliu0820 committed 2 years ago

11d4d91b

Add loss operators to fp32 cast policy of AutocastCPU (#81689)

CaoE committed 2 years ago

a09c3fcb

[PT-D][BE][TP perf 1/N] Get rid of unnecessary collectives in Embedding/EmbeddingBag and use autograd-enabled collectives (#81853)

fduwjj committed 2 years ago

d3a176a1

Avoid using true division in split_dim (#83527)

ezyang committed 2 years ago

e09821f7

NestedTensor Softmax (#83435)

mikekgfb committed 2 years ago

d7fc76a1

[TorchTidy] Adding support for accessing strides and scalars (#80072)

Gamrix committed 2 years ago

343b5f86

Fix `torch.equal` on CPU (#83350)

malfet committed 2 years ago

1a09b05c

add the nessesairy constraints for the next 5 benchmarks (#82923)

migeed-z committed 2 years ago

df62ea76

Optionally run fbgemm in tracer (#83531)

JacobSzwejbka committed 2 years ago

aac622ad

[MPS] Fix conv1D and conv2D with non-matching strides/paddings (#83522)

kulinseth committed 2 years ago

31d4b6f5

use global var for disabled and slow test dicts (#83487)

clee2000 committed 2 years ago

0e2efaf9

Older