strisunshinewentingwang/pytorch

Pull Requests Commits

Emulate weight and activation only quant with fake quant, numerics test (#26625)

raghuramank100 committed 4 years ago

bc3fb36e

Quantization aware training: Freeze batch norm support (#26624)

raghuramank100 committed 4 years ago

e0822f10

Per channel fake quant (#26623)

raghuramank100 committed 4 years ago

cbfd4e05

Improve repr for quantized modules

James Reed committed 4 years ago

b9a2c8ac

Per-channel baseline (#26516)

raghuramank100 committed 4 years ago

6d7a73c0

Dont zero out buffers in dynamic linear (#27002)

James Reed committed 4 years ago

15e48276

fix AvgPool2d for 2^31-1 sized inputs, and get test_cuda_kernel_loop_overflow_large to working state

soumith committed 4 years ago

024fa347

fix segfault when printing error msg for list comp (#27398)

Krovatkin committed 4 years ago

95d2c7fc

Make `align_to` method-only. (#27304) (#27367)

zou3519 committed 4 years ago

7ba2baee

add AutoNonVariableTypeMode for USE_STATIC_DISPATCH on JIT->ATen path (#27274) (#27321)

ljk53 committed 4 years ago

ccf3a6de

Fixed Error message for tensor.align_to (#27221) (#27250)

izdeby committed 4 years ago

1ba6fc4c

Enabled comparison ops with named tensors (#27162) (#27249)

izdeby committed 4 years ago

d4d4bf56

Fix ONNX Interpolate (#27233)

houseroad committed 4 years ago

cee965fa

make class types callable (#26743) (#27226)

eellison committed 4 years ago

544c16cd

[jit] Fix `toIValue` dict iteration (#27112)

driazati committed 4 years ago

494a5563

Module method destroy (#27111)

IvanKobzarev committed 4 years ago

ba4c3a1c

Bump gloo (#27087)

pietern committed 4 years ago

c556f905

Turn on named tensor testing for v1.3.0 (#27084)

zou3519 committed 4 years ago

5c80dd3c

[jit] Serializing autograd ops into its own namespace (#27079)

wanchaol committed 4 years ago

2d8ee111

Serialize XLA Tensor (#27042)

ailzhang committed 4 years ago

ebc2519b

make cudnn rnn respect current stream (#27044)

ngimel committed 4 years ago

8c9e4b25

fix pytorch_linux_xenial_py3_6_gcc5_4_build for release branch

soumith committed 4 years ago

6a6f047f

Update to ROCm 2.8 (#27337)

bddppq committed 4 years ago

deadc27c

xta0 committed 4 years ago

6276fda1

Provide (but skip) 3.5 job by default on all PRs. (#27293)

ezyang committed 4 years ago

65ee8f2c

Report docker push / pull time (#26861)

ezyang committed 4 years ago

6126cfab

Don't apply should_run to the nightly/postnightly branches. (#27061)

ezyang committed 4 years ago

e2f6fed6

Turn Caffe2 CUDA 9.1 + py2 to CUDA 10.1 + py3 (#26835)

houseroad committed 4 years ago

667deb92

fix OSX CI build (#27373)

soumith committed 4 years ago

0e88de55

Make nonzero non differentiable as it supposed to be (#26980)

VitalyFedyunin committed 4 years ago

3c8ce2a5

Newer Older