strisunshinewentingwang/pytorch

Pull Requests Commits

Introduce the torch.distributed.ops module

cbalioglu committed 3 years ago

290ea8a6

LU solve uses cuBLAS and cuSOLVER for matrices with dim > 1024 (#61815)

rkansal47 committed 3 years ago

6d21e36f

[sharded_tensor] add default fields to ShardedTensorMetadata (#62867)

wanchaol committed 3 years ago

0c39cea3

[DDP] Dont set thread local state in reducer autograd hook. (#62996)

rohan-varma committed 3 years ago

5fb79f61

[typing] suppress errors in `fbcode/caffe2` - batch 2

Pyre Bot Jr committed 3 years ago

6915bc07

Test shape analysis with opinfos (#59814)

Elias Ellison committed 3 years ago

ea808df2

add ssupport for a few more opinfos in jit (#59812)

Elias Ellison committed 3 years ago

7312bd95

Don't substitute in symbolic shapes to shape compute graph (#59811)

Elias Ellison committed 3 years ago

9cbdc90d

small cleanups (#59810)

Elias Ellison committed 3 years ago

7db0bcfb

Only optimize after change (redo) (#59809)

Elias Ellison committed 3 years ago

9cd990de

[jit] warn if _check_overload_body fails to find source

suo committed 3 years ago

4c630773

[quant] Update get_default_qat_qconfig to return the fused observer+fake_quant module (#62702)

supriyar committed 3 years ago

aa89d5f7

[quant] add reduce_range option to FusedMovingAvgFakeQuantize module (#62863)

supriyar committed 3 years ago

08d1a12d

Codegen: Fix operator::name on windows (#62278)

peterbell10 committed 3 years ago

978490d7

Reject kwonly arguments passed positionally in torch.ops (#62981)

ezyang committed 3 years ago

cdf702b6

Allow LocalResponseNorm to accept 0 dim batch sizes (#62801)

v0dro committed 3 years ago

9e7b6bb6

Update TensorPipe submodule

lw committed 3 years ago

061062ae

[Reland][DDP] Support not all outputs used in loss calculation (#61753)

rohan-varma committed 3 years ago

3df48703

To fix variance computation for complex Adam (#62946)

iramazanli committed 3 years ago

5ed6e442

[quant][graphmode][fx] Attach a weight qparam dict to linear and conv in reference quantized model (#62488)

jerryzh168 committed 3 years ago

3c1d1170

Simplify the logic of running ci workflow codegen (#62853)

zhouzhuojie committed 3 years ago

59ac451b

irange-ify 12b (#62484)

r-barnes committed 3 years ago

8720369a

Shard Operators.cpp (#62185)

peterbell10 committed 3 years ago

93e0f3a3

irange-ify 13d (#62477)

r-barnes committed 3 years ago

4b9ca72c

Enable rebuilds for Ninja on Windows (#62948)

peterjc123 committed 3 years ago

d16587f8

BFP16 quantization/dequantization (#62974)

marjanfariborz committed 3 years ago

a82b9ef1

Migrate Embedding thrust sort to cub sort (#62495)

zasdfgbnm committed 3 years ago

c4aeecac

Use output memory format based on input for cudnn_convolution_relu (#62482)

= committed 3 years ago

084e92bb

irange-ify 12 (#62120)

r-barnes committed 3 years ago

4fdb9579

irange-ify 1 (#62193)

r-barnes committed 3 years ago

da9958c8

Older