strisunshinewentingwang/pytorch

Pull Requests Commits

Revert D16238204: Revert D16224780: [pytorch][PR] [ROCm] MIOpen integration into pytorch RNN operators

bddppq committed 5 years ago

92468f0a

Revert D16037021: Support GRU module quantization in Pytorch

Karl Ostmo committed 5 years ago

da315a4e

Revert D16224780: [pytorch][PR] [ROCm] MIOpen integration into pytorch RNN operators

Karl Ostmo committed 5 years ago

fcfefc34

Transfer Learning: Caffe2 load op changes to return shape inference (#22829)

pradeepdfb committed 5 years ago

ead11932

Support GRU module quantization in Pytorch

lly-zero-one committed 5 years ago

d8c1b861

Get rid of torch.mean shape analysis

James Reed committed 5 years ago

ba9d559a

Use Linear Operator with fp16 weights in JIT (#22323)

mingzhe09088 committed 5 years ago

9eb03933

Support Linear operation with fp16 weights in ATen (#22023)

mingzhe09088 committed 5 years ago

573d9e69

Revert D16204820: [pytorch][PR] optimize topk on cpu using parallel and partial sort

Karl Ostmo committed 5 years ago

35ee4bf4

add support for breaks and continues (#21692)

Elias Ellison committed 5 years ago

cf2889ad

PyTorch export to ONNX Opset 7 and 8 - Cont (#22421)

BowenBao committed 5 years ago

b3147bc6

MIOpen integration into pytorch RNN operators (#22774)

ashish committed 5 years ago

9f8e2c06

Speeds up fast-path for 1D tensors (#22756)

colesbury committed 5 years ago

30e03df6

avoid kernel launches for zero-sized tensor inputs

wanchaol committed 5 years ago

02bc06a6

Make PythonArgs::tensor and PythonArgs::scalar faster (#22782)

colesbury committed 5 years ago

b1b65f34

optimize topk on cpu using parallel and partial sort (#19736)

mingfeima committed 5 years ago

10c14ad1

Speed up TensorIterator::compute_strides a little (#22779)

colesbury committed 5 years ago

fc23d7f3

Initiate checkCuda90Bug warning (#22757)

Guanheng Zhang committed 5 years ago

f266a63e

Revert D16222539: [pytorch][PR] Let users pass CMake-specific options starting with CMAKE_ to CMake.

ezyang committed 5 years ago

ccb28939

Let users pass CMake-specific options starting with CMAKE_ to CMake. (#22776)

xuhdev committed 5 years ago

612eed31

add new tests to benchmark_all_test (#22787)

mingzhe09088 committed 5 years ago

7eb03193

make custom op work in OSS environment (#22781)

mingzhe09088 committed 5 years ago

1878800f

Add the support of handle Bias being nullptr for torch.ops.quantized.fbgemm_linear (#22403)

jianyuh committed 5 years ago

8ec712da

Update ONNX constant folding to support opset 10. (#22515)

Spandan Tiwari committed 5 years ago

9d11004e

make CompilationUnit::define return defined functions

suo committed 5 years ago

291570e0

refactor self to be a class again

suo committed 5 years ago

de819be9

Give functions qualified names

suo committed 5 years ago

22d70e0d

Suppress progress bar only for pip install

ezyang committed 5 years ago

4b48ae4a

Remove hard-coded NVRTC specific constant from fuser header

bwasti committed 5 years ago

05d56bd1

assert_no_internal_overlap pass op name by const ref

James Reed committed 5 years ago

513b7a7a

Newer Older