microsoft
/
nni
зеркало из https://github.com/microsoft/nni.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
2 232 коммитов 79 Ветки 49 Теги 420 MiB
Граф коммитов

2232 Коммитов

Автор SHA1 Сообщение Дата
Zhenhua Han f2f58dbb55
[Retiarii] cross-graph optimization: device placement and input deduplication (#3202) 2021-07-30 19:11:02 +08:00
QuanluZhang 6645bd332d
fix multi-thread issue (#3899) 2021-07-30 09:33:25 +08:00
QuanluZhang 703f6de58c
Fix pipeline (#3994) 2021-07-29 23:13:23 +08:00
SparkSnail b42f85f5e1
Fix kubeflow & frameworkcontroller pipeline (#3991) 2021-07-29 14:44:46 +08:00
Bill Wu aa2cc9223e
Transformer Head Pruner (#3884) 2021-07-28 09:02:45 +08:00
chenbohua3 370e88dfd4
Add post training observer_quantizer (#3915) 2021-07-28 09:01:46 +08:00
J-shang 3f1e4f55b7
[Model Compression] fix compression doc and update a notebook example (#3955) 2021-07-28 09:01:01 +08:00
Ningxin Zheng 30361a2e87
update compression speedup documentation (#3979) 2021-07-27 17:14:54 +08:00
J-shang 726a46de7c
fix bug (#3977) 2021-07-27 17:14:21 +08:00
Yuge Zhang a7846135a8
Enable `fixed_arch` on Retiarii (#3972) 2021-07-27 17:14:01 +08:00
Yuge Zhang 08fe2924f5
Merge pull request #3938 from microsoft/nn-meter 
[DO NOT SQUASH] Support nn-Meter in Retiarii framework
 2021-07-27 13:51:26 +08:00
Yuge Zhang 5e04d56c51
Replace example_inputs with dummy_input (#3983) 2021-07-26 22:17:21 +08:00
Zhenhua Han 3bce6926ac
extend reusable training service to support placement constraint (#3897) 2021-07-26 19:30:04 +08:00
liuzhe-lz b0de7c9326
Add search space validation for choice types (#3975) 2021-07-26 15:08:26 +08:00
SparkSnail ef9e27b9b3
Support kubeflow reuse mode (#3919) 2021-07-26 14:15:40 +08:00
J-shang 322424ab7f
Add nnimanager ut back (#3970) 2021-07-26 10:34:17 +08:00
Bill Wu 68818a3385
[Model Compression] Expand export_model arguments: dummy input and onnx opset_version (#3968) 2021-07-26 10:23:43 +08:00
liuzhe-lz deef0c424b
Update docs to v2 config (#3966) 2021-07-26 10:16:43 +08:00
thomasschmied 03a02232ba
Change .to() to to_device() (#3963) 2021-07-26 10:12:45 +08:00
Bill Wu 5bf2cb192a
HPO Benchmark Fixes and New Features (#3925) 2021-07-26 10:09:16 +08:00
chenbohua3 7fc5af0702
Add batch normalization folding to QAT quantizer (#3911) 2021-07-26 09:56:02 +08:00
Yuge Zhang 441c5da542
Extend windows test timeout (#3962) 2021-07-23 17:14:09 +08:00
Yuge Zhang 7d101f8371
Use python 3.9 on pipeline (#3881) 2021-07-23 17:08:21 +08:00
liuzhe-lz f27b874190
JupyterLab extension (#3954) 2021-07-23 11:32:12 +08:00
J-shang 9270f9b807
Update confidence=8 in speed up test (#3967) 2021-07-21 14:13:10 +08:00
Ningxin Zheng b65830e0c7
Support UnSqueeze (#3960) 2021-07-21 12:36:41 +08:00
liuzhe-lz 442342cb19
Fix bug in IP detection (#3952) 2021-07-21 10:17:41 +08:00
kalineid bb3974807e
Add tests for GraphConverterWithShape (#3951) 2021-07-20 14:24:18 +08:00
lin bin d8127e0284
[Model Compression] Add global sort for taylor pruner (#3896) 2021-07-20 14:23:51 +08:00
Ningxin Zheng c80ed3e904
Remove dumplicated log message. (#3950) 2021-07-19 16:01:54 +08:00
QuanluZhang d765e51194
remove useless file (#3949) 2021-07-17 10:02:29 +08:00
lin bin 1d9caa61ea
[Quantization Speedup] Unit Test (#3862) 2021-07-16 17:32:05 +08:00
kvartet 7205cff578
Update Python API doc (#3885) 2021-07-16 17:31:32 +08:00
Ni Hao c1873c5fe9
add nfs shared storage integration test (#3913) 2021-07-16 17:28:28 +08:00
liuzhe-lz 094ed9f172
"Fix" NNI manager UT (#3933) 2021-07-16 17:22:39 +08:00
Mark 5588adea39
Allow for environment variable to set NNI configuration directory (#3936) 
Co-authored-by: Mark Janse <m.h.a.janse-2@umcutrecht.nl>
 2021-07-16 17:22:23 +08:00
Zhenhua Han bccfa9fff1
execute removeGpuReservation first to release GPUs properly (#3941) 2021-07-16 17:22:07 +08:00
Ningxin Zheng 30df69bf5b
Fix integration test on speedup examples (#3945) 2021-07-16 14:06:58 +08:00
Yuge Zhang 403195f0eb
Merge branch 'master' into nn-meter 2021-07-15 16:02:47 +08:00
kalineid 99aa8226c6
[Retiarii]: Add info required by nn-meter to graph ir (#3910) 
* Fix mutable default

* LayerChoice:forward now will default run the first candidate to support trace (#3910)

* New GraphConverter to parse shape info required by nn-meter (#3910)

* Support model filter in Random strategy

* Support latency aware search in SPOS multi-trial example

* Fix for review (#3910)

* Add doc for hardware-aware NAS

* Fix lint python & Add nn_meter to sphinx mock

* Add comments

* Move LatencyFilter to examples

* Move example inputs into configs

* Support nested layer choice

Co-authored-by: Jianyu Wei <v-wjiany@microsoft.com>
Co-authored-by: kalineid <nnob@mail.ustc.edu.cn>
Co-authored-by: Yuge Zhang <scottyugochang@gmail.com>
Co-authored-by: Yuge Zhang <Yuge.Zhang@microsoft.com>
 2021-07-15 16:01:16 +08:00
Lijiaoa a7278d2d25
fix search broken bug with incomplete filter (#3935) 2021-07-15 15:44:46 +08:00
QuanluZhang 5eb43ea781
[Retiarii] support hypermodule: autoactivation (#3868) 2021-07-15 13:42:28 +08:00
Ningxin Zheng 47c7ea14f1
Add several speedup examples (#3880) 2021-07-15 11:41:39 +08:00
QuanluZhang 5fe245006a
[retiarii] support torch 1.8 and 1.9 (#3937) 2021-07-14 17:15:23 +08:00
Yuge Zhang 542a660d05
[Retiarii] NAS-Bench-201 (#3920) 2021-07-14 14:30:12 +08:00
Ningxin Zheng 7eedec46aa
Model Speedup Refactor (#3462) 2021-07-14 14:24:40 +08:00
Yuge Zhang 5b99b5984c
Fix pipeline dependency (#3926) 2021-07-14 12:51:03 +08:00
liuzhe-lz 0c3827b3a5
fix ip detection (#3934) 
Co-authored-by: liuzhe <zhe.liu@microsoft.com>
 2021-07-14 11:29:55 +08:00
Bill Wu 994a2d2363
[Example] Conditional GAN in pytorch (#3687) 2021-07-14 11:17:37 +08:00
J-shang 68644f5963
add exclude config validate in compressor (#3815) 2021-07-13 07:46:33 +08:00