strisunshinewentingwang/pytorch

Pull Requests Commits

Add USE_NAMEDTENSOR compilation flag. (#20162)

zou3519 committed 5 years ago

e01a5bf2

Fix in file position logic: file descriptor and Python-side handle (#20270)

philippslang committed 5 years ago

f23fb66e

error instead of crashing on attempt to subclass typed tensors (#20283)

nairbv committed 5 years ago

c406bf20

Switch off USE_DISTRIBUTED on default for MSVC (#20302)

peterjc123 committed 5 years ago

1e35ef07

Fixed Softmax doc to specify dimension to prevent warning in 1.1.0. (#20310)

ZaydH committed 5 years ago

2aea5b63

Use torch::get/set_num_threads without additional includes beyond torch/torch.h (#20176)

Will Feng committed 5 years ago

0087069d

Sort User Defined Classes (#19706)

eellison committed 5 years ago

916ef768

Bump op_version_set (#19812)

zdevito committed 5 years ago

1102f4c5

Refactor ChunkDataReader API + fix missing headers (#19485)

Thiago Crepaldi committed 5 years ago

3d4d7b90

Eliminate some const_cast's.

resistor committed 5 years ago

bed1d7d3

Only record grad_fn in C++ Scatter and Gather when required so (#20286)

mrshenli committed 5 years ago

d6815e1e

Dynamic quantized full LSTM module (#20249)

James Reed committed 5 years ago

2179d5b3

Enable ONNX constant folding in test_pytorch_onnx_caffe2.py tests. (#20290)

Spandan Tiwari committed 5 years ago

7bc8562a

ScaleBlobs Operator (#19660)

bilgeacun committed 5 years ago

3ee97183

split canonicalize_ops, make a decompose pass (#19988)

wanchaol committed 5 years ago

4d676d53

dont make alias for none value (#20112)

Elias Ellison committed 5 years ago

33b4afe3

Support `torch.save` for saving values during execution (#18154)

davidriazati committed 5 years ago

8ebb86dd

Unify code path for native mkldnn conv and reorder on-the-fly mkldnn conv (#19963)

bddppq committed 5 years ago

27c82c03

Have add_video use NamedTemporaryFile directly (#20223)

lanpa committed 5 years ago

b3bce01e

Canonicalize order of If and Loop outputs (#20015)

Elias Ellison committed 5 years ago

35de90e3

Disable worker_kill & holder_iter_reference combination in test_proper_exit (#20172)

ssnl committed 5 years ago

1ab33fce

Clarify API and add examples for all methods (#20008)

lanpa committed 5 years ago

7edf9a25

Quantized FCRelu operator (#19750)

jianyuh committed 5 years ago

4a086f70

Use QTensor with quantized FC operator (#19541)

jianyuh committed 5 years ago

e8cdfb5d

Enable caffe2 softmax tests with ROCm 2.4 (#20280)

bddppq committed 5 years ago

8defcbfc

Quantized FC operator (#19497)

jianyuh committed 5 years ago

1deb8dde

Rebase conflict fix for isFusableDevice (#20251)

bwasti committed 5 years ago

7b733e4f

SubgraphRewriter: Add a support for arbitrary replacement graphs in subgraph rewriter. (#20084)

Mikhail Zolotukhin committed 5 years ago

c931d7e9

SubgraphRewriter: Expose runOnGraph and use it in tests. (#20083)

Mikhail Zolotukhin committed 5 years ago

b3324d0f

SubgraphRewriter: Rename pattern fusion to subgraph rewrite. (#20082)

Mikhail Zolotukhin committed 5 years ago

8a6072c3

Newer Older