strisunshinewentingwang/pytorch

Pull Requests Commits

Make validate private by renaming validate to _validate (#107927)

gmagogsfm committed 1 year ago

39854df1

Add tests for foreach copy (#107860)

mlazos committed 1 year ago

f2f82855

[core][sparse][pruning] cuSPARSELt Kernels and ops. (#107398)

cpuhrsch committed 1 year ago

925d71e7

Enhance fakepg: add fsdp+tp tests (#107626)

xmfan committed 1 year ago

c2ac0da4

improved error message for IO mismatch (#107907)

avikchaudhuri committed 1 year ago

bfcd2645

Expose torch.export.{save,load} APIs (#107888)

gmagogsfm committed 1 year ago

bfb09204

add get buffer from exported program (#107809)

cccclai committed 1 year ago

4f2ff1d0

[quant][pt2e] Make sure XNNPACKQuantizer works with the pre_dispatch=True (#107872)

jerryzh168 committed 1 year ago

a0cfaf06

Avoid decomposing `_unsafe_index` in Inductor (#107882)

SS-JIA committed 1 year ago

86f9fec3

Fix the example of torch.slice_scatter (#107849)

FFFrog committed 1 year ago

e00bd831

[inductor][ac] preserve recompute tags through pattern matching (#107742)

anijain2305 committed 1 year ago

8b7b824d

[vision hash update] update the pinned vision hash (#107911)

pytorchupdatebot committed 1 year ago

df2ca187

[MPS] Implement `polar` via metal shader (#107324)

malfet committed 1 year ago

6e85a688

[ONNX] Enable 'ExportOutput.save' for models larger than 2GB (#107904)

BowenBao committed 1 year ago

00e9735e

[inductor][fx passes] batch tanh in pre grad (#107881)

jackiexu1992 committed 1 year ago

fc33dc01

Expose cudaStreamCaptureMode in CUDA Graphs, use local setting in inductor (#107407)

eellison committed 1 year ago

0a9778a3

profiler tree test: skip cudaGetDeviceProperties_v2, cudaGetDeviceCount (#107887)

davidberard98 committed 1 year ago

c18d2a3c

[FSDP] verify backward_prefetch works correctly with unit test (#107058)

weifengpy committed 1 year ago

ec10b17c

Improve unbacked symint error msg (#107806)

tugsbayasgalan committed 1 year ago

485de730

[1/N] fix clang-tidy warnings in torch/csrc (#107648)

cyyever committed 1 year ago

49eeca00

Expose ExportedProgram and related classes (#107852)

gmagogsfm committed 1 year ago

7dd11134

[c10d] Increase socket buffer size to allow ProcessGroup init up to 12k ranks (#107878)

XilunWu committed 1 year ago

49fbaa29

[PyTorch][Tensor] Introduce tensor.dim_order (#106835)

digantdesai committed 1 year ago

8a7a6867

[pytorch][Quant] Fix bias quant bug (#107810)

kimishpatel committed 1 year ago

2fbe6ef2

[aot_inductor] fix hardcoded output dtype (#107825)

frank-wei committed 1 year ago

497571df

Enhance fakepg: send and recv (#107625)

xmfan committed 1 year ago

870fa460

[Inductor][MacOS] resolve macos openmp problem and provide a holistic instruction (#107111)

youkaichao committed 1 year ago

66dc1aba

[Dynamo] cache_size policy (#107496)

anijain2305 committed 1 year ago

4175a6e9

Add SingletonSymIntNode (#107089)

soulitzer committed 1 year ago

d7130e97

Update nccl submodule to 2.18.5 (#107883)

Skylion007 committed 1 year ago

a41d15e4

Newer Older