strisunshinewentingwang/pytorch

Pull Requests Commits

Update references to minimum CUDA and cuDNN version (#20718)

colesbury committed 5 years ago

6ec3c122

CUDA implementation of fakequant (#20252)

jerryzh168 committed 5 years ago

05543153

Revert D15445092: Some minor fix to unblock the Bert model quantization

ezyang committed 5 years ago

fdb92399

fix add_histogram_raw (#20688)

lanpa committed 5 years ago

cfc98ae7

Exposing LengthsSum/Mean/Max in pytorch (#20802)

kittipatv committed 5 years ago

fd2aa93b

ONNX Export Scatter

lara-hdr committed 5 years ago

8d7a0257

Add ability to filter metric schema in LayerModelHelper (#20786)

shunting314 committed 5 years ago

fea4a56a

Automatic update of fbcode/onnx to cc2333a3f929caca7223b98699237f19388dd585 (#20763)

houseroad committed 5 years ago

810816a1

Fix optimizer type hint (#20648)

TheCodez committed 5 years ago

4e0d098a

When detecting numpy, assign relavant variables outside the try block (#20739)

xuhdev committed 5 years ago

795a1a6f

Revert D15248618: Split ATen/Parallel into interface and backend

ezyang committed 5 years ago

fd95947e

Some minor fix to unblock the Bert model quantization (#20787)

lly-zero-one committed 5 years ago

70ecddfd

Add quant-dequant nodes for bias. (#20045)

nishantpdce committed 5 years ago

a501e7d5

Add DictType to Metadata (#20770)

houseroad committed 5 years ago

c2d0e731

Use AT_INTERNAL_ASSERT in test_base (#20555)

suo committed 5 years ago

70eb315d

Rename FC to Linear in the test routine (#20716)

jianyuh committed 5 years ago

4a85e795

fbgemm precision argument (#20790)

jerryzh168 committed 5 years ago

77651615

Split ATen/Parallel into interface and backend (#20057)

Ilia Cherniavskii committed 5 years ago

c4a3b4d5

int_repr for different quantized types (#20656)

jerryzh168 committed 5 years ago

adbab828

Use SmallVector to allocate Compound operands inline. (#20762)

Owen Anderson committed 5 years ago

c1d6bcf3

Optimize pytorch layer_norm forward (#20345)

xiaomengy committed 5 years ago

c9da0119

use tensoriterator instead of th for fill_ implementation. (#20719)

nairbv committed 5 years ago

9cec8ae1

Fix copy_transpose_valid check (#20759)

colesbury committed 5 years ago

7a0c6d52

make magic methods work with casts too (#20654)

Elias Ellison committed 5 years ago

5acc664f

Change Bias to QTensor with qint32(int32_t) (#20713)

jianyuh committed 5 years ago

e6f22e1b

Change Weight to QTensor with qint8(int8_t) (#20712)

jianyuh committed 5 years ago

b9a150ed

Fix a bug in quantize_linear (#20711)

jianyuh committed 5 years ago

ac2314fd

Update Conda description in PyTorch README (#20726)

dreiss committed 5 years ago

32803b52

Auto-convert GPU arrays that support the __cuda_array_interface__ protocol (#20584)

madsbk committed 5 years ago

5d8879cf

Improve the recommended citation

xuhdev committed 5 years ago

847d9c57

Newer Older