strisunshinewentingwang/pytorch

Pull Requests Commits

[quant][fx] Add RNN reference module (#73386)

jerryzh168 committed 2 years ago

81437e66

Make debug_pkl smaller by only emitting unique traces. (#73368)

qihqi committed 2 years ago

61d6c438

Eliminate unused parameters in PyTorch (#73625)

r-barnes committed 2 years ago

07df8876

[quant] Add support for multiple inputs in fusion pattern (#73572)

jerryzh168 committed 2 years ago

bea075f3

[Static Runtime] Fix a broken test & Add an out variant wrapper for `mse_loss` (#73574)

d1jang committed 2 years ago

539acb29

[cuDNN v8 API] cuDNN benchmark, convolution bwd / transposed convolution fwd, `bfloat16`, conv-bias-activation fusion (#60755)

eqy committed 2 years ago

dab5e2a2

[pkg] improve error message for module detection on saving pass (#73106)

PaliC committed 2 years ago

e7051939

[ROCM] Navi21 Enablement 4: Normalization kernels (#73543)

micmelesse committed 2 years ago

7b8fc748

Deletes unused line in test_autocast_rnn (#73195)

mcarilli committed 2 years ago

2f957f51

only run complex autograd tests once

albanD committed 2 years ago

701fa16e

simplify run_test for distributed tests

albanD committed 2 years ago

f275b3f9

parse TernaryIf as right associative, fix #68221 (#68416)

etern committed 2 years ago

dc81ba1f

Write explicit meta-kernels for `normal` (#70089)

nkaretnikov committed 2 years ago

eb0d370f

Unify checks for `normal` (#70087)

nkaretnikov committed 2 years ago

dad0e0c3

Fix a typo: add a missing space (#70086)

nkaretnikov committed 2 years ago

08493e03

Bug fix: allow std 0 in the meta definition of `normal_` (#70085)

nkaretnikov committed 2 years ago

e9c1ccee

[Quant][core][gpu] Implemented support for relu in quantized conv operator using cudnn (#73337)

dzdang committed 2 years ago

19572950

[Quant][core][gpu] Implemented support for bias in quantized conv operator in cudnn (#73035)

dzdang committed 2 years ago

4e54c1cc

[PyTorch] Add codegen unboxing ability (#69881)

larryliu0820 committed 2 years ago

9ce9803a

[FSDP] Make summon_full_params a public method (#73116)

fegin committed 2 years ago

6396547f

[Reland] Add BUILD_LAZY_CUDA_LINALG option (#73447)

malfet committed 2 years ago

6302cdb9

ci: Migrate aws creds to managed IAM

seemethere committed 2 years ago

ebf0ca33

Disable forward/backward correlation to workaround the profiler crash (#72904)

guotuofeng committed 2 years ago

554169fc

'typename Base' is checked repeatedly (#72842)

MambaWong committed 2 years ago

f42202d2

Improve logging on ONNX symbolic helpers

Thiago Crepaldi committed 2 years ago

6643522d

(torch/elastic) skip logging structured error info if error_file is not set (#73477)

Kiuk Chung committed 2 years ago

b08309ee

Remove AutoHeaders.RECURSIVE_GLOB from caffe2/ (#73227)

Marc Fisher committed 2 years ago

9e60b003

Reserve the memory for vector to save cost in `gather_ranges_to_dense_op.h` (#73540)

Su Shi committed 2 years ago

1c621a7b

Revert D33767740: [pytorch][PR] Sparse CSR CPU: cuSolverSP backend for `linalg.solve`

malfet committed 2 years ago

8ac73935

Use set -ex instead of many &&s

janeyx99 committed 2 years ago

97cf4c6b

Newer Older