strisunshinewentingwang/pytorch

Pull Requests Commits

[jit] Enable type tags in serialization (#35741)

davidriazati committed 4 years ago

596153ca

[RPC][Better Engineering] Consolidated all rpcAgentRunning atomic booleans (#33915)

osalpekar committed 4 years ago

19bbfbe1

Add quick utility to transform scripted/traced models for mobile. (#35904)

kimishpatel committed 4 years ago

c5c63a2e

Set eval mode during optimization for mobile. (#35903)

kimishpatel committed 4 years ago

f48008c2

[jit] Fix type comparisons segfault (#35929)

davidriazati committed 4 years ago

6e13a778

Refactor C++ API parity test mechanism and turn it on in CI again (#35190)

Will Feng (FAIAR) committed 4 years ago

2fa3c157

Remove python2 and 3.5 from requirements.txt, README and docs (#35677)

orionr committed 4 years ago

2d8dbcd3

[quant][graphmode] Add quantize_per_tensor.tensors (#35916)

supriyar committed 4 years ago

7468ef04

[quant][graphmode] Insert Observers for dynamic LSTM (#35894)

supriyar committed 4 years ago

f0c74724

[quant][graphmode] Add new tensorlist observer for LSTM (#35893)

supriyar committed 4 years ago

0429d2c9

Make RRef type_hint mismatch exception message more actionable to users (#35943)

xush6528 committed 4 years ago

87582ae6

Make intdiv_256 a more generic binary operator template (#35422)

xuhdev committed 4 years ago

ea8021d7

lshift and rshift on CUDA should match the behavior on CPU (#35339)

xuhdev committed 4 years ago

a1cf3fd1

Make intdiv_256 a more generic binary operator template (#35422)

xuhdev committed 4 years ago

beac3f27

Fix gcc-5.4 compilation (#35935)

malfet committed 4 years ago

e707cee5

[ROCm] [ROCm 2.10+] enable fp16 dot in Caffe2 backend (#30432)

iotamudelta committed 4 years ago

be125d18

Add `GRAPH_UPDATE` for `x.size()` in Peephole Optimize (#34865)

invalid-email-address committed 4 years ago

8484ca58

Make ValType hashable. (#35917)

csarofeen committed 4 years ago

aeb13f21

Vec256<bfloat16>::arange step size should accept templates. (#35842)

xuhdev committed 4 years ago

1a146b05

Use full include path in autogenerated Functions.cpp (#35924)

malfet committed 4 years ago

a5af478f

Avoid one unnecessary memory allocation in XNNPACK integration. (#35350)

Ashkan Aliabadi committed 4 years ago

d0ce94d2

Fixes a bug in serializing min/max plus one more. (#35850)

kimishpatel committed 4 years ago

c33ea41f

Report CUDA separate compilation flag (#35726)

malfet committed 4 years ago

e2adcc1c

Clear profiling information timely and appropriately (#35814)

Krovatkin committed 4 years ago

767ea03b

.circleci: Bump libtorch builds to 3.7 (#35912)

seemethere committed 4 years ago

1a723269

Removes integer division call sites (#35862)

Mike Ruberry committed 4 years ago

591b5da2

[PTM] register aten::dequantize.self for spark spot int8 model

linbinyu committed 4 years ago

ced1e463

[easy] ThroughputBenchmark: make ScriptModuleBenchmark usable from c++ (#35848)

Alexander Sidorov committed 4 years ago

f5b95748

add c10d dynamic loading mechanism and unit test (#28068)

ftian1 committed 4 years ago

762270c5

Fix contant/placeholder loading in checkGraphCompatibility

jackm321 committed 4 years ago

2a4ca708

Newer Older