strisunshinewentingwang/pytorch

Pull Requests Commits

[quant] Enable per-channel quantization for quantized nn.LSTM

supriyar committed 4 years ago

3724d3f0

[quant] Make reduce_range default for RNNs

supriyar committed 4 years ago

8f317515

Add %= support in TorchScript (#38983)

ShawnZhong committed 4 years ago

f872cf5e

Revert D21769463: [pytorch][PR] Refactor c10::complex and cleanup c10::Scalar

albanD committed 4 years ago

8556664d

Refactor c10::complex and cleanup c10::Scalar (#38593)

zasdfgbnm committed 4 years ago

928ce29e

[AMD] Fix broken test (#39297)

xw285cornell committed 4 years ago

fcef4396

[ONNX] Remove Aten ops from ONNX export (#37239)

neginraoof committed 4 years ago

b7b99ab0

[nnpi fake ops] bug fix int8QuantizeNNPI (#39271)

amylittleyang committed 4 years ago

c02cb7aa

work around building onnx in older rocm docker images (#39253)

jeffdaily committed 4 years ago

b0d6e4b6

[quant] Dynamic Linear module to use reduce_range (#39125)

supriyar committed 4 years ago

25a6c5f6

[quant] Add reduce_range argument for qlinear_dynamic (#39041)

supriyar committed 4 years ago

9cacbe29

Avoid a TensorIterator/Loops reinterpret_cast in a test. (#39246)

gchanan committed 4 years ago

001102c5

Remove some unnecessary code in Onnxifi (#39197)

Yinghai Lu committed 4 years ago

7ab96461

Set RecordFunction id only when needed (#39265)

Ilia Cherniavskii committed 4 years ago

a5e023f2

test_fc_nnpi_fp16.py test_fc_num0_fix fix (#39248)

venkatacrc committed 4 years ago

1c67c3d5

[quant][graphmode] Rename _quantize_script.py to quantize_script.py (#39122)

jerryzh168 committed 4 years ago

1d0ec50a

Added real and imag views as tensor attributes (#39033)

anjali411 committed 4 years ago

a50d781c

Fix init-shutdown race condition in autograd engine (#39194)

malfet committed 4 years ago

c3d3782c

[nnpi eval] enable int8 eval with emulation Int8FC (#39112)

amylittleyang committed 4 years ago

88c5fd94

Followup for cuda assert cleanups (#39220)

peterjc123 committed 4 years ago

29c04acd

Implement CUDA_KERNEL_ASSERT for MSVC (#39218)

peterjc123 committed 4 years ago

a5d44800

[ROCm] in test_cuda.py, re-enable skipped tests (#37952)

jeffdaily committed 4 years ago

c25b3d43

[quant][graphmode] Cleanup inplace API (#38827)

jerryzh168 committed 4 years ago

85d0292c

[JIT] JIT should let people know we inferred an argument as a tensor (#38527)

xuewenc committed 4 years ago

7836eace

Fix assertNotEqual error reporting (#39217)

malfet committed 4 years ago

86f46ac9

Update NNPI backend to v0.6.0.5 (#4539)

omromano committed 4 years ago

f44fca88

Fix windows upload jobs (#39249)

peterjc123 committed 4 years ago

b44f02f8

support complex types for `cumsum`, `cumprod` (#39063)

kshitij12345 committed 4 years ago

10e2126b

Revert D21751663: [pytorch][PR] Fix argmin/max bug

ngimel committed 4 years ago

4b5e87f9

Improve warnings to actually point at user code (#39143)

albanD committed 4 years ago

d6715e63

Older