strisunshinewentingwang/pytorch

Pull Requests Commits

[quant] Move EmbeddingBag eager quantization to static (#44217)

supriyar committed 4 years ago

646ffd48

[quant] Add quantized Embedding module (#44208)

supriyar committed 4 years ago

57b87aaf

[quant] Support quantization of embedding lookup operators (#44207)

supriyar committed 4 years ago

6013a29f

[caffe2] fix wrong comment (#42735)

jspark1105 committed 4 years ago

f27be2f7

fix lint (#44346)

Elias Ellison committed 4 years ago

f9146b45

[quant][graphmode][fx][api] Call fuse in prepare (#43984)

jerryzh168 committed 4 years ago

6269b6e0

[NNC] fix support for FP16 in CudaCodgen (#44209)

nickgg committed 4 years ago

be94dba4

[quant][graphmode][fx] Support inplace option (#43983)

jerryzh168 committed 4 years ago

9f54bcc5

add rocm nightly build (#44250)

Rong Rong committed 4 years ago

0351d317

Added alpha overloads for add/sub ops with lists (#43413)

Iurii Zdebskyi committed 4 years ago

40d138f7

fx quant: add docblocks to _find_matches and _find_quants (#43928)

vkuzo committed 4 years ago

00b5bd53

[fix] output of `embedding_bag` with non-contiguous weight (#44032)

kshitij12345 committed 4 years ago

6dd53fb5

[quant][graphmode][fx] Support quantize per channel in all cases (#44042)

jerryzh168 committed 4 years ago

43e38d60

Set default compiler differently according to platform (#43890)

scintiller committed 4 years ago

49e979bf

[FX] Minor fixups in Graph printout (#44214)

James Reed committed 4 years ago

1fcccd6a

Enable temp disabled tests in test_jit_fuser_te.py (#44222)

Krovatkin committed 4 years ago

47ac9bb1

Release saved variable from DifferentiableGraphBackward (#42994)

sujoysaraswati committed 4 years ago

54931ebb

Skips test_addcmul_cuda if using ROCm (#44304)

Mike Ruberry committed 4 years ago

63d62d3e

Reduce `sccache` log levels for RocM to a default state (#44310)

malfet committed 4 years ago

de89261a

Don't register a fallback for private use to let extensions do it themselves (#44149)

ezyang committed 4 years ago

477f4891

[JIT] Unshare types for modules that define() in __init__ (#44233)

Meghan Lele committed 4 years ago

caf23d11

[FX] Only copy over training attr if it\'s there (#44314)

James Reed committed 4 years ago

4e0ac120

quant: switch observers to use min_max (#42957)

vkuzo committed 4 years ago

fd8e2064

skip test_tanhquantize for now (#44312)

hl475 committed 4 years ago

de980f93

add 'run_duration' stats for binary builds to scuba (#44251)

Rong Rong committed 4 years ago

8d212d3f

Automated submodule update: FBGEMM (#44177)

facebook-github-bot committed 4 years ago

1130de79

[dper3] Export Caffe2 operator LearningRate to PyTorch

Brandon Lin committed 4 years ago

5de805d8

Add unary ops: exp and sqrt (#42537)

Iurii Zdebskyi committed 4 years ago

cce5982c

Revert D23561500: Benchmarks: re-enable profiling-te configuration (try 2).

Mikhail Zolotukhin committed 4 years ago

6134ac17

test_ops: skipTest only takes a single argument (#44181)

t-vi committed 4 years ago

7c61f57b

Newer Older