{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":529844475,"defaultBranch":"master","name":"mmrazor","ownerLogin":"aptsunny","currentUserCanPush":false,"isFork":true,"isEmpty":false,"createdAt":"2022-08-28T11:39:35.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/36404164?v=4","public":true,"private":false,"isOrgOwned":false},"refInfo":{"name":"","listCacheKey":"v0:1682140898.0","currentOid":""},"activityList":{"items":[{"before":null,"after":"91663815fee310cc19e4e502089569a48ed438af","ref":"refs/heads/main","pushedAt":"2023-04-22T05:21:38.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"aptsunny","name":"Yue Sun","path":"/aptsunny","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/36404164?s=80&v=4"},"commit":{"message":"[Feature] Add MMRazor quantization (#513)\n\n* [FEATURE] add quant algo `Learned Step Size Quantization` (#346)\r\n\r\n* update\r\n\r\n* Fix a bug in make_divisible. (#333)\r\n\r\nfix bug in make_divisible\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* [Fix] Fix counter mapping bug (#331)\r\n\r\n* fix counter mapping bug\r\n\r\n* move judgment into get_counter_type & update UT\r\n\r\n* [Docs]Add MMYOLO projects link (#334)\r\n\r\n* [Doc] fix typos in en/usr_guides (#299)\r\n\r\n* Update README.md\r\n\r\n* Update README_zh-CN.md\r\n\r\nCo-authored-by: Sheffield <49406546+SheffieldCao@users.noreply.github.com>\r\n\r\n* [Features]Support `MethodInputsRecorder` and `FunctionInputsRecorder` (#320)\r\n\r\n* support MethodInputsRecorder and FunctionInputsRecorder\r\n\r\n* fix bugs that the model can not be pickled\r\n\r\n* WIP: add pytest for ema model\r\n\r\n* fix bugs in recorder and delivery when ema_hook is used\r\n\r\n* don't register the DummyDataset\r\n\r\n* fix pytest\r\n\r\n* updated\r\n\r\n* retina loss & predict & tesnor DONE\r\n\r\n* [Feature] Add deit-base (#332)\r\n\r\n* WIP: support deit\r\n\r\n* WIP: add deithead\r\n\r\n* WIP: fix checkpoint hook\r\n\r\n* fix data preprocessor\r\n\r\n* fix cfg\r\n\r\n* WIP: add readme\r\n\r\n* reset single_teacher_distill\r\n\r\n* add metafile\r\n\r\n* add model to model-index\r\n\r\n* fix configs and readme\r\n\r\n* [Feature]Feature map visualization (#293)\r\n\r\n* WIP: vis\r\n\r\n* WIP: add visualization\r\n\r\n* WIP: add visualization hook\r\n\r\n* WIP: support razor visualizer\r\n\r\n* WIP\r\n\r\n* WIP: wrap draw_featmap\r\n\r\n* support feature map visualization\r\n\r\n* add a demo image for visualization\r\n\r\n* fix typos\r\n\r\n* change eps to 1e-6\r\n\r\n* add pytest for visualization\r\n\r\n* fix vis hook\r\n\r\n* fix arguments' name\r\n\r\n* fix img path\r\n\r\n* support draw inference results\r\n\r\n* add visualization doc\r\n\r\n* fix figure url\r\n\r\n* move files\r\n\r\nCo-authored-by: weihan cao <HIT-cwh>\r\n\r\n* [Feature] Add kd examples (#305)\r\n\r\n* support kd for mbv2 and shufflenetv2\r\n\r\n* WIP: fix ckpt path\r\n\r\n* WIP: fix kd r34-r18\r\n\r\n* add metafile\r\n\r\n* fix metafile\r\n\r\n* delete\r\n\r\n* [Doc] add documents about pruning. (#313)\r\n\r\n* init\r\n\r\n* update user guide\r\n\r\n* update images\r\n\r\n* update\r\n\r\n* update How to prune your model\r\n\r\n* update how_to_use_config_tool_of_pruning.md\r\n\r\n* update doc\r\n\r\n* move location\r\n\r\n* update\r\n\r\n* update\r\n\r\n* update\r\n\r\n* add mutablechannels.md\r\n\r\n* add references\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: jacky <jacky@xx.com>\r\n\r\n* [Feature] PyTorch version of `PKD: General Distillation Framework for Object Detectors via Pearson Correlation Coefficient`. (#304)\r\n\r\n* add pkd\r\n\r\n* add pytest for pkd\r\n\r\n* fix cfg\r\n\r\n* WIP: support fcos3d\r\n\r\n* WIP: support fcos3d pkd\r\n\r\n* support mmdet3d\r\n\r\n* fix cfgs\r\n\r\n* change eps to 1e-6 and add some comments\r\n\r\n* fix docstring\r\n\r\n* fix cfg\r\n\r\n* add assert\r\n\r\n* add type hint\r\n\r\n* WIP: add readme and metafile\r\n\r\n* fix readme\r\n\r\n* update metafiles and readme\r\n\r\n* fix metafile\r\n\r\n* fix pipeline figure\r\n\r\n* for RFC\r\n\r\n* Customed FX initialize\r\n\r\n* add UT init\r\n\r\n* [Refactor] Refactor Mutables and Mutators (#324)\r\n\r\n* refactor mutables\r\n\r\n* update load fix subnet\r\n\r\n* add DumpChosen Typehint\r\n\r\n* adapt UTs\r\n\r\n* fix lint\r\n\r\n* Add GroupMixin to ChannelMutator (temporarily)\r\n\r\n* fix type hints\r\n\r\n* add GroupMixin doc-string\r\n\r\n* modified by comments\r\n\r\n* fix type hits\r\n\r\n* update subnet format\r\n\r\n* fix channel group bugs and add UTs\r\n\r\n* fix doc string\r\n\r\n* fix comments\r\n\r\n* refactor diff module forward\r\n\r\n* fix error in channel mutator doc\r\n\r\n* fix comments\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* [Fix] Update readme (#341)\r\n\r\n* update kl readme\r\n\r\n* update dsnas readme\r\n\r\n* fix url\r\n\r\n* Bump version to 1.0.0rc1 (#338)\r\n\r\nupdate version\r\n\r\n* init demo\r\n\r\n* add customer_tracer\r\n\r\n* add quantizer\r\n\r\n* add fake_quant, loop, config\r\n\r\n* remove CPatcher in custome_tracer\r\n\r\n* demo_try\r\n\r\n* init version\r\n\r\n* modified base.py\r\n\r\n* pre-rebase\r\n\r\n* wip of adaround series\r\n\r\n* adaround experiment\r\n\r\n* trasfer to s2\r\n\r\n* update api\r\n\r\n* point at sub_reconstruction\r\n\r\n* pre-checkout\r\n\r\n* export onnx\r\n\r\n* add customtracer\r\n\r\n* fix lint\r\n\r\n* move custom tracer\r\n\r\n* fix import\r\n\r\n* TDO: UTs\r\n\r\n* Successfully RUN\r\n\r\n* update loop\r\n\r\n* update loop docstrings\r\n\r\n* update quantizer docstrings\r\n\r\n* update qscheme docstrings\r\n\r\n* update qobserver docstrings\r\n\r\n* update tracer docstrings\r\n\r\n* update UTs init\r\n\r\n* update UTs init\r\n\r\n* fix review comments\r\n\r\n* fix CI\r\n\r\n* fix UTs\r\n\r\n* update torch requirements\r\n\r\nCo-authored-by: huangpengsheng <huangpengsheng@sensetime.com>\r\nCo-authored-by: LKJacky <108643365+LKJacky@users.noreply.github.com>\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: Yang Gao <Gary1546308416AL@gmail.com>\r\nCo-authored-by: kitecats <90194592+kitecats@users.noreply.github.com>\r\nCo-authored-by: Sheffield <49406546+SheffieldCao@users.noreply.github.com>\r\nCo-authored-by: whcao <41630003+HIT-cwh@users.noreply.github.com>\r\nCo-authored-by: jacky <jacky@xx.com>\r\nCo-authored-by: pppppM <67539920+pppppM@users.noreply.github.com>\r\nCo-authored-by: humu789 <humu@pjlab.org.cn>\r\n\r\n* [Features]Quantize pipeline (#350)\r\n\r\n* init demo\r\n\r\n* add customer_tracer\r\n\r\n* add quantizer\r\n\r\n* add fake_quant, loop, config\r\n\r\n* remove CPatcher in custome_tracer\r\n\r\n* demo_try\r\n\r\n* init version\r\n\r\n* modified base.py\r\n\r\n* pre-rebase\r\n\r\n* wip of adaround series\r\n\r\n* adaround experiment\r\n\r\n* trasfer to s2\r\n\r\n* update api\r\n\r\n* point at sub_reconstruction\r\n\r\n* pre-checkout\r\n\r\n* export onnx\r\n\r\n* add customtracer\r\n\r\n* fix lint\r\n\r\n* move custom tracer\r\n\r\n* fix import\r\n\r\n* update\r\n\r\n* updated\r\n\r\n* retina loss & predict & tesnor DONE\r\n\r\n* for RFC\r\n\r\n* Customed FX initialize\r\n\r\n* add UT init\r\n\r\n* TDO: UTs\r\n\r\n* Successfully RUN\r\n\r\n* update loop\r\n\r\n* update loop docstrings\r\n\r\n* update quantizer docstrings\r\n\r\n* update qscheme docstrings\r\n\r\n* update qobserver docstrings\r\n\r\n* update tracer docstrings\r\n\r\n* update UTs init\r\n\r\n* update UTs init\r\n\r\n* fix bugs\r\n\r\n* fix lsq\r\n\r\n* refactor quantize pipeline\r\n\r\n* fix quant\r\n\r\n* WIP: debug qat\r\n\r\n* fix lsq bugs\r\n\r\n* fix qat, docstring in progress\r\n\r\n* TDO: UTs\r\n\r\n* fix bugs\r\n\r\n* fix lsq\r\n\r\n* refactor quantize pipeline\r\n\r\n* fix quant\r\n\r\n* WIP: debug qat\r\n\r\n* fix lsq bugs\r\n\r\n* fix qat, docstring in progress\r\n\r\n* fixed DefaultQconfigs name\r\n\r\n* fix bugs\r\n\r\n* add comments and fix typos\r\n\r\n* delete useless codes\r\n\r\n* fix bugs and add comments\r\n\r\n* rename prepare_module_dict\r\n\r\n* update lsq config\r\n\r\nCo-authored-by: humu789 <humu@pjlab.org.cn>\r\nCo-authored-by: huangpengsheng <huangpengsheng@sensetime.com>\r\nCo-authored-by: FreakieHuang <frank0huang@foxmail.com>\r\nCo-authored-by: pppppM <gjf_mail@126.com>\r\n\r\n* [Feature] Add `prepare_for_mmdeploy` interface  (#365)\r\n\r\n* remove useless code\r\n\r\n* fix build graph module import bug\r\n\r\n* refactor general quant\r\n\r\n* rename GeneralQuant to MMArchitectureQuant\r\n\r\n* fix some dtype bugs\r\n\r\n* add prepare_for_mmdeploy interface\r\n\r\n* update prepare for mmdeploy args\r\n\r\n* fix some comments\r\n\r\nCo-authored-by: humu789 <humu@pjlab.org.cn>\r\n\r\n* CodeCamp #132 add MinMaxFloorObserver (#376)\r\n\r\n* add minmaxfloor_observer.py\r\n\r\n* add MinMaxFloorObserver and normative docstring\r\n\r\n* add test for MinMaxFloorObserver\r\n\r\n* Quant go (#409)\r\n\r\n* add torch observer\r\n\r\n* add torch fakequant\r\n\r\n* refactor base quantizer\r\n\r\n* add QConfigHander and QSchemeHander & finish quantizer_refactor_beta\r\n\r\n* passed ptq_pipeline\r\n\r\n* tmp-commit\r\n\r\n* fix loop and algorithm\r\n\r\n* delete fakequant\r\n\r\n* refactor code structure\r\n\r\n* remove lsq\r\n\r\n* valid ptq pipeline\r\n\r\n* wip\r\n\r\n* fix del functions\r\n\r\n* fix\r\n\r\n* fix lint and pytest\r\n\r\nCo-authored-by: HIT-cwh <2892770585@qq.com>\r\n\r\n* [Refactor & Doc] Refactor graph_utils and add docstring and pytest (#420)\r\n\r\n* refactor graph_utils and add docstring and pytest\r\n\r\n* fix del fakequant\r\n\r\n* delete useless codes\r\n\r\n* Merge dev-1.x into quantize (#430)\r\n\r\n* Fix a bug in make_divisible. (#333)\r\n\r\nfix bug in make_divisible\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* [Fix] Fix counter mapping bug (#331)\r\n\r\n* fix counter mapping bug\r\n\r\n* move judgment into get_counter_type & update UT\r\n\r\n* [Docs]Add MMYOLO projects link (#334)\r\n\r\n* [Doc] fix typos in en/usr_guides (#299)\r\n\r\n* Update README.md\r\n\r\n* Update README_zh-CN.md\r\n\r\nCo-authored-by: Sheffield <49406546+SheffieldCao@users.noreply.github.com>\r\n\r\n* [Features]Support `MethodInputsRecorder` and `FunctionInputsRecorder` (#320)\r\n\r\n* support MethodInputsRecorder and FunctionInputsRecorder\r\n\r\n* fix bugs that the model can not be pickled\r\n\r\n* WIP: add pytest for ema model\r\n\r\n* fix bugs in recorder and delivery when ema_hook is used\r\n\r\n* don't register the DummyDataset\r\n\r\n* fix pytest\r\n\r\n* [Feature] Add deit-base (#332)\r\n\r\n* WIP: support deit\r\n\r\n* WIP: add deithead\r\n\r\n* WIP: fix checkpoint hook\r\n\r\n* fix data preprocessor\r\n\r\n* fix cfg\r\n\r\n* WIP: add readme\r\n\r\n* reset single_teacher_distill\r\n\r\n* add metafile\r\n\r\n* add model to model-index\r\n\r\n* fix configs and readme\r\n\r\n* [Feature]Feature map visualization (#293)\r\n\r\n* WIP: vis\r\n\r\n* WIP: add visualization\r\n\r\n* WIP: add visualization hook\r\n\r\n* WIP: support razor visualizer\r\n\r\n* WIP\r\n\r\n* WIP: wrap draw_featmap\r\n\r\n* support feature map visualization\r\n\r\n* add a demo image for visualization\r\n\r\n* fix typos\r\n\r\n* change eps to 1e-6\r\n\r\n* add pytest for visualization\r\n\r\n* fix vis hook\r\n\r\n* fix arguments' name\r\n\r\n* fix img path\r\n\r\n* support draw inference results\r\n\r\n* add visualization doc\r\n\r\n* fix figure url\r\n\r\n* move files\r\n\r\nCo-authored-by: weihan cao <HIT-cwh>\r\n\r\n* [Feature] Add kd examples (#305)\r\n\r\n* support kd for mbv2 and shufflenetv2\r\n\r\n* WIP: fix ckpt path\r\n\r\n* WIP: fix kd r34-r18\r\n\r\n* add metafile\r\n\r\n* fix metafile\r\n\r\n* delete\r\n\r\n* [Doc] add documents about pruning. (#313)\r\n\r\n* init\r\n\r\n* update user guide\r\n\r\n* update images\r\n\r\n* update\r\n\r\n* update How to prune your model\r\n\r\n* update how_to_use_config_tool_of_pruning.md\r\n\r\n* update doc\r\n\r\n* move location\r\n\r\n* update\r\n\r\n* update\r\n\r\n* update\r\n\r\n* add mutablechannels.md\r\n\r\n* add references\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: jacky <jacky@xx.com>\r\n\r\n* [Feature] PyTorch version of `PKD: General Distillation Framework for Object Detectors via Pearson Correlation Coefficient`. (#304)\r\n\r\n* add pkd\r\n\r\n* add pytest for pkd\r\n\r\n* fix cfg\r\n\r\n* WIP: support fcos3d\r\n\r\n* WIP: support fcos3d pkd\r\n\r\n* support mmdet3d\r\n\r\n* fix cfgs\r\n\r\n* change eps to 1e-6 and add some comments\r\n\r\n* fix docstring\r\n\r\n* fix cfg\r\n\r\n* add assert\r\n\r\n* add type hint\r\n\r\n* WIP: add readme and metafile\r\n\r\n* fix readme\r\n\r\n* update metafiles and readme\r\n\r\n* fix metafile\r\n\r\n* fix pipeline figure\r\n\r\n* [Refactor] Refactor Mutables and Mutators (#324)\r\n\r\n* refactor mutables\r\n\r\n* update load fix subnet\r\n\r\n* add DumpChosen Typehint\r\n\r\n* adapt UTs\r\n\r\n* fix lint\r\n\r\n* Add GroupMixin to ChannelMutator (temporarily)\r\n\r\n* fix type hints\r\n\r\n* add GroupMixin doc-string\r\n\r\n* modified by comments\r\n\r\n* fix type hits\r\n\r\n* update subnet format\r\n\r\n* fix channel group bugs and add UTs\r\n\r\n* fix doc string\r\n\r\n* fix comments\r\n\r\n* refactor diff module forward\r\n\r\n* fix error in channel mutator doc\r\n\r\n* fix comments\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* [Fix] Update readme (#341)\r\n\r\n* update kl readme\r\n\r\n* update dsnas readme\r\n\r\n* fix url\r\n\r\n* Bump version to 1.0.0rc1 (#338)\r\n\r\nupdate version\r\n\r\n* [Feature] Add Autoformer algorithm (#315)\r\n\r\n* update candidates\r\n\r\n* update subnet_sampler_loop\r\n\r\n* update candidate\r\n\r\n* add readme\r\n\r\n* rename variable\r\n\r\n* rename variable\r\n\r\n* clean\r\n\r\n* update\r\n\r\n* add doc string\r\n\r\n* Revert \"[Improvement] Support for candidate multiple dimensional search constraints.\"\r\n\r\n* [Improvement] Update Candidate with multi-dim search constraints. (#322)\r\n\r\n* update doc\r\n\r\n* add support type\r\n\r\n* clean code\r\n\r\n* update candidates\r\n\r\n* clean\r\n\r\n* xx\r\n\r\n* set_resource -> set_score\r\n\r\n* fix ci bug\r\n\r\n* py36 lint\r\n\r\n* fix bug\r\n\r\n* fix check constrain\r\n\r\n* py36 ci\r\n\r\n* redesign candidate\r\n\r\n* fix pre-commit\r\n\r\n* update cfg\r\n\r\n* add build_resource_estimator\r\n\r\n* fix ci bug\r\n\r\n* remove runner.epoch in testcase\r\n\r\n* [Feature] Autoformer architecture and dynamicOPs (#327)\r\n\r\n* add DynamicSequential\r\n\r\n* dynamiclayernorm\r\n\r\n* add dynamic_pathchembed\r\n\r\n* add DynamicMultiheadAttention and DynamicRelativePosition2D\r\n\r\n* add channel-level dynamicOP\r\n\r\n* add autoformer algo\r\n\r\n* clean notes\r\n\r\n* adapt channel_mutator\r\n\r\n* vit fly\r\n\r\n* fix import\r\n\r\n* mutable init\r\n\r\n* remove annotation\r\n\r\n* add DynamicInputResizer\r\n\r\n* add unittest for mutables\r\n\r\n* add OneShotMutableChannelUnit_VIT\r\n\r\n* clean code\r\n\r\n* reset unit for vit\r\n\r\n* remove attr\r\n\r\n* add autoformer backbone UT\r\n\r\n* add valuemutator UT\r\n\r\n* clean code\r\n\r\n* add autoformer algo UT\r\n\r\n* update classifier UT\r\n\r\n* fix test error\r\n\r\n* ignore\r\n\r\n* make lint\r\n\r\n* update\r\n\r\n* fix lint\r\n\r\n* mutable_attrs\r\n\r\n* fix test\r\n\r\n* fix error\r\n\r\n* remove DynamicInputResizer\r\n\r\n* fix test ci\r\n\r\n* remove InputResizer\r\n\r\n* rename variables\r\n\r\n* modify type\r\n\r\n* Continued improvements of ChannelUnit\r\n\r\n* fix lint\r\n\r\n* fix lint\r\n\r\n* remove OneShotMutableChannelUnit\r\n\r\n* adjust derived type\r\n\r\n* combination mixins\r\n\r\n* clean code\r\n\r\n* fix sample subnet\r\n\r\n* search loop fly\r\n\r\n* more annotations\r\n\r\n* avoid counter warning and modify batch_augment cfg by gy\r\n\r\n* restore\r\n\r\n* source_value_mutables restriction\r\n\r\n* simply arch_setting api\r\n\r\n* update\r\n\r\n* clean\r\n\r\n* fix ut\r\n\r\n* [Feature] Add performance predictor (#306)\r\n\r\n* add predictor with 4 handlers\r\n\r\n* [Improvement] Update Candidate with multi-dim search constraints. (#322)\r\n\r\n* update doc\r\n\r\n* add support type\r\n\r\n* clean code\r\n\r\n* update candidates\r\n\r\n* clean\r\n\r\n* xx\r\n\r\n* set_resource -> set_score\r\n\r\n* fix ci bug\r\n\r\n* py36 lint\r\n\r\n* fix bug\r\n\r\n* fix check constrain\r\n\r\n* py36 ci\r\n\r\n* redesign candidate\r\n\r\n* fix pre-commit\r\n\r\n* update cfg\r\n\r\n* add build_resource_estimator\r\n\r\n* fix ci bug\r\n\r\n* remove runner.epoch in testcase\r\n\r\n* update metric_predictor:\r\n1. update MetricPredictor;\r\n2. add predictor config for searching;\r\n3. add predictor in evolution_search_loop.\r\n\r\n* add UT for predictor\r\n\r\n* add MLPHandler\r\n\r\n* patch optional.txt for predictors\r\n\r\n* patch test_evolution_search_loop\r\n\r\n* refactor apis of predictor and handlers\r\n\r\n* fix ut and remove predictor_cfg in predictor\r\n\r\n* adapt new mutable & mutator design\r\n\r\n* fix ut\r\n\r\n* remove unness assert after rebase\r\n\r\n* move predictor-build in __init__ & simplify estimator-build\r\n\r\nCo-authored-by: Yue Sun <aptsunny@tongji.edu.cn>\r\n\r\n* [Feature] Add DCFF (#295)\r\n\r\n* add ChannelGroup (#250)\r\n\r\n* rebase new dev-1.x\r\n\r\n* modification for adding config_template\r\n\r\n* add docstring to channel_group.py\r\n\r\n* add docstring to mutable_channel_group.py\r\n\r\n* rm channel_group_cfg from Graph2ChannelGroups\r\n\r\n* change choice type of SequentialChannelGroup from float to int\r\n\r\n* add a warning about group-wise conv\r\n\r\n* restore __init__ of dynamic op\r\n\r\n* in_channel_mutable  ->  mutable_in_channel\r\n\r\n* rm abstractproperty\r\n\r\n* add a comment about VT\r\n\r\n* rm registry for ChannelGroup\r\n\r\n* MUTABLECHANNELGROUP -> ChannelGroupType\r\n\r\n* refine docstring of IndexDict\r\n\r\n* update docstring\r\n\r\n* update docstring\r\n\r\n* is_prunable -> is_mutable\r\n\r\n* update docstring\r\n\r\n* fix error in pre-commit\r\n\r\n* update unittest\r\n\r\n* add return type\r\n\r\n* unify init_xxx apit\r\n\r\n* add unitest about init of MutableChannelGroup\r\n\r\n* update according to reviews\r\n\r\n* sequential_channel_group -> sequential_mutable_channel_group\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* Add BaseChannelMutator and refactor Autoslim (#289)\r\n\r\n* add BaseChannelMutator\r\n\r\n* add autoslim\r\n\r\n* tmp\r\n\r\n* make SequentialMutableChannelGroup accpeted both of num and ratio  as choice. and supports divisior\r\n\r\n* update OneShotMutableChannelGroup\r\n\r\n* pass supernet training of autoslim\r\n\r\n* refine autoslim\r\n\r\n* fix bug in OneShotMutableChannelGroup\r\n\r\n* refactor make_divisible\r\n\r\n* fix spell error:  channl -> channel\r\n\r\n* init_using_backward_tracer -> init_from_backward_tracer\r\ninit_from_fx_tracer -> init_from_fx_tracer\r\n\r\n* refine SequentialMutableChannelGroup\r\n\r\n* let mutator support models with dynamicop\r\n\r\n* support define search space in model\r\n\r\n* tracer_cfg -> parse_cfg\r\n\r\n* refine\r\n\r\n* using -> from\r\n\r\n* update docstring\r\n\r\n* update docstring\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* tmpsave\r\n\r\n* migrate ut\r\n\r\n* tmpsave2\r\n\r\n* add loss collector\r\n\r\n* refactor slimmable and add l1-norm (#291)\r\n\r\n* refactor slimmable and add l1-norm\r\n\r\n* make l1-norm support convnd\r\n\r\n* update get_channel_groups\r\n\r\n* add  l1-norm_resnet34_8xb32_in1k.py\r\n\r\n* add pretrained to resnet34-l1\r\n\r\n* remove old channel mutator\r\n\r\n* BaseChannelMutator -> ChannelMutator\r\n\r\n* update according to reviews\r\n\r\n* add readme to l1-norm\r\n\r\n* MBV2_slimmable -> MBV2_slimmable_config\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* update config\r\n\r\n* fix md & pytorch support <1.9.0 in batchnorm init\r\n\r\n* Clean old codes. (#296)\r\n\r\n* remove old dynamic ops\r\n\r\n* move dynamic ops\r\n\r\n* clean old mutable_channels\r\n\r\n* rm OneShotMutableChannel\r\n\r\n* rm MutableChannel\r\n\r\n* refine\r\n\r\n* refine\r\n\r\n* use SquentialMutableChannel to replace OneshotMutableChannel\r\n\r\n* refactor dynamicops folder\r\n\r\n* let SquentialMutableChannel support float\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* fix ci\r\n\r\n* ci fix py3.6.x & add mmpose\r\n\r\n* ci fix py3.6.9 in utils/index_dict.py\r\n\r\n* fix mmpose\r\n\r\n* minimum_version_cpu=3.7\r\n\r\n* fix ci 3.7.13\r\n\r\n* fix pruning &meta ci\r\n\r\n* support python3.6.9\r\n\r\n* fix py3.6 import caused by circular import patch in py3.7\r\n\r\n* fix py3.6.9\r\n\r\n* Add channel-flow (#301)\r\n\r\n* base_channel_mutator -> channel_mutator\r\n\r\n* init\r\n\r\n* update docstring\r\n\r\n* allow omitting redundant configs for channel\r\n\r\n* add register_mutable_channel_to_a_module to MutableChannelContainer\r\n\r\n* update according to reviews 1\r\n\r\n* update according to reviews 2\r\n\r\n* update according to reviews 3\r\n\r\n* remove old docstring\r\n\r\n* fix error\r\n\r\n* using->from\r\n\r\n* update according to reviews\r\n\r\n* support self-define input channel number\r\n\r\n* update docstring\r\n\r\n* chanenl -> channel_elem\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: jacky <jacky@xx.com>\r\n\r\n* support >=3.7\r\n\r\n* support py3.6.9\r\n\r\n* Rename: ChannelGroup -> ChannelUnit (#302)\r\n\r\n* refine repr of MutableChannelGroup\r\n\r\n* rename folder name\r\n\r\n* ChannelGroup -> ChannelUnit\r\n\r\n* filename in units folder\r\n\r\n* channel_group -> channel_unit\r\n\r\n* groups -> units\r\n\r\n* group -> unit\r\n\r\n* update\r\n\r\n* get_mutable_channel_groups -> get_mutable_channel_units\r\n\r\n* fix bug\r\n\r\n* refine docstring\r\n\r\n* fix ci\r\n\r\n* fix bug in tracer\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\n\r\n* update new channel config format\r\n\r\n* update pruning refactor\r\n\r\n* update merged pruning\r\n\r\n* update commit\r\n\r\n* fix dynamic_conv_mixin\r\n\r\n* update comments: readme&dynamic_conv_mixins.py\r\n\r\n* update readme\r\n\r\n* move kl softmax channel pooling to op by comments\r\n\r\n* fix comments: fix redundant & split README.md\r\n\r\n* dcff in ItePruneAlgorithm\r\n\r\n* partial dynamic params for fuseconv\r\n\r\n* add step_freq & prune_time check\r\n\r\n* update comments\r\n\r\n* update comments\r\n\r\n* update comments\r\n\r\n* fix ut\r\n\r\n* fix gpu ut & revise step_freq in ItePruneAlgorithm\r\n\r\n* update readme\r\n\r\n* revise ItePruneAlgorithm\r\n\r\n* fix docs\r\n\r\n* fix dynamic_conv attr\r\n\r\n* fix ci\r\n\r\nCo-authored-by: LKJacky <108643365+LKJacky@users.noreply.github.com>\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: zengyi.vendor <zengyi.vendor@sensetime.com>\r\nCo-authored-by: jacky <jacky@xx.com>\r\n\r\n* [Fix] Fix optional requirements (#357)\r\n\r\n* fix optional requirements\r\n\r\n* fix dcff ut\r\n\r\n* fix import with get_placeholder\r\n\r\n* supplement the previous commit\r\n\r\n* [Fix] Fix configs of wrn models and ofd. (#361)\r\n\r\n* 1.revise the configs of wrn22, wrn24, and wrn40. 2.revise the data_preprocessor of ofd_backbone_resnet50_resnet18_8xb16_cifar10\r\n\r\n* 1.Add README for vanilla-wrm.\r\n\r\n* 1.Revise readme of wrn\r\n\r\nCo-authored-by: zhangzhongyu <zhangzhongyu@pjlab.org.cn>\r\n\r\n* [Fix] Fix bug on mmrazor visualization, mismatch argument in define and use. (#356)\r\n\r\nfix bug on mmrazor visualization, mismatch argument in define and use.\r\n\r\nCo-authored-by: Xianpan Zhou <32625100+PanDaMeow@users.noreply.github.com>\r\n\r\n* fix bug in benchmark_test (#364)\r\n\r\nfix bug in configs\r\n\r\nCo-authored-by: Your Name <you@example.com>\r\n\r\n* [FIX] Fix wrn configs (#368)\r\n\r\n* fix wrn configs\r\n\r\n* fix wrn configs\r\n\r\n* update online wrn model weight\r\n\r\n* [Fix] fix bug on pkd config. Wrong import filename. (#373)\r\n\r\n* [CI] Update ci to torch1.13 (#380)\r\n\r\nupdate ci to torch1.13\r\n\r\n* [Feature] Add BigNAS algorithm (#219)\r\n\r\n* add calibrate-bn-statistics\r\n\r\n* add test calibrate-bn-statistics\r\n\r\n* fix mixins\r\n\r\n* fix mixins\r\n\r\n* fix mixin tests\r\n\r\n* remove slimmable channel mutable and refactor dynamic op\r\n\r\n* refact dynamic batch norm\r\n\r\n* add progressive dynamic conv2d\r\n\r\n* add center crop dynamic conv2d\r\n\r\n* refactor dynamic directory\r\n\r\n* refactor dynamic sequential\r\n\r\n* rename length to depth in dynamic sequential\r\n\r\n* add test for derived mutable\r\n\r\n* refactor dynamic op\r\n\r\n* refactor api of dynamic op\r\n\r\n* add derive mutable mixin\r\n\r\n* addbignas algorithm\r\n\r\n* refactor bignas structure\r\n\r\n* add input resizer\r\n\r\n* add input resizer to bignas\r\n\r\n* move input resizer from algorithm into classifier\r\n\r\n* remove compnents\r\n\r\n* add attentive mobilenet\r\n\r\n* delete json file\r\n\r\n* nearly(less 0.2) align inference accuracy with gml\r\n\r\n* move mutate seperated in bignas mobilenet backbone\r\n\r\n* add zero_init_residual\r\n\r\n* add set_dropout\r\n\r\n* set dropout in bignas algorithm\r\n\r\n* fix registry\r\n\r\n* add subnet yaml and nearly align inference accuracy with gml\r\n\r\n* add rsb config for bignas\r\n\r\n* remove base in config\r\n\r\n* add gml bignas config\r\n\r\n* convert to iter based\r\n\r\n* bignas forward and backward fly\r\n\r\n* fix merge conflict\r\n\r\n* fix dynamicseq bug\r\n\r\n* fix bug and refactor bignas\r\n\r\n* arrange configs of bignas\r\n\r\n* fix typo\r\n\r\n* refactor attentive_mobilenet\r\n\r\n* fix channel mismatch due to registion of DerivedMutable\r\n\r\n* update bignas & fix se channel mismatch\r\n\r\n* add AutoAugmentV2 & remove unness configs\r\n\r\n* fix lint\r\n\r\n* recover channel assertion in channel unit\r\n\r\n* fix a group bug\r\n\r\n* fix comments\r\n\r\n* add docstring\r\n\r\n* add norm in dynamic_embed\r\n\r\n* fix search loop & other minor changes\r\n\r\n* fix se expansion\r\n\r\n* minor change\r\n\r\n* add ut for bignas & attentive_mobilenet\r\n\r\n* fix ut\r\n\r\n* update bignas readme\r\n\r\n* rm unness ut & supplement get_placeholder\r\n\r\n* fix lint\r\n\r\n* fix ut\r\n\r\n* add subnet deployment in downstream tasks.\r\n\r\n* minor change\r\n\r\n* update ofa backbone\r\n\r\n* minor fix\r\n\r\n* Continued improvements of searchable backbone\r\n\r\n* minor change\r\n\r\n* drop ratio in backbone\r\n\r\n* fix comments\r\n\r\n* fix ci test\r\n\r\n* fix test\r\n\r\n* add dynamic shortcut UT\r\n\r\n* modify strategy to fit bignas\r\n\r\n* fix test\r\n\r\n* fix bug in neck\r\n\r\n* fix error\r\n\r\n* fix error\r\n\r\n* fix yaml\r\n\r\n* save subnet ckpt\r\n\r\n* merge autoslim_val/test_loop into subnet_val_loop\r\n\r\n* move calibrate_bn_mixin to utils\r\n\r\n* fix bugs and add docstring\r\n\r\n* clean code\r\n\r\n* fix register bug\r\n\r\n* clean code\r\n\r\n* update\r\n\r\nCo-authored-by: wangshiguang <wangshiguang@sensetime.com>\r\nCo-authored-by: gaoyang07 <1546308416@qq.com>\r\nCo-authored-by: aptsunny <aptsunny@tongji.edu.cn>\r\nCo-authored-by: sunyue1 <sunyue1@sensetime.com>\r\n\r\n* [Bug] Fix ckpt (#372)\r\n\r\nfix ckpt\r\n\r\n* [Feature] Add tools to convert distill ckpt to student-only ckpt. (#381)\r\n\r\n* [Feature] Add tools to convert distill ckpt to student-only ckpt.\r\n\r\n* fix bug.\r\n\r\n* add --model-only to only save model.\r\n\r\n* Make changes accroding to PR review.\r\n\r\n* Enhance the Abilities of the Tracer for Pruning. (#371)\r\n\r\n* tmp\r\n\r\n* add new mmdet models\r\n\r\n* add docstring\r\n\r\n* pass test and pre-commit\r\n\r\n* rm razor tracer\r\n\r\n* update fx tracer, now it can automatically wrap methods and functions.\r\n\r\n* update tracer passed models\r\n\r\n* add warning for torch <1.12.0\r\n\r\nfix bug for python3.6\r\n\r\nupdate placeholder to support placeholder.XXX\r\n\r\n* fix bug\r\n\r\n* update docs\r\n\r\n* fix lint\r\n\r\n* fix parse_cfg in configs\r\n\r\n* restore mutablechannel\r\n\r\n* test ite prune algorithm when using dist\r\n\r\n* add get_model_from_path to MMModelLibrrary\r\n\r\n* add mm models to DefaultModelLibrary\r\n\r\n* add uts\r\n\r\n* fix bug\r\n\r\n* fix bug\r\n\r\n* add uts\r\n\r\n* add uts\r\n\r\n* add uts\r\n\r\n* add uts\r\n\r\n* fix bug\r\n\r\n* restore ite_prune_algorithm\r\n\r\n* update doc\r\n\r\n* PruneTracer -> ChannelAnalyzer\r\n\r\n* prune_tracer -> channel_analyzer\r\n\r\n* add test for fxtracer\r\n\r\n* fix bug\r\n\r\n* fix bug\r\n\r\n* PruneTracer -> ChannelAnalyzer\r\n\r\nrefine\r\n\r\n* CustomFxTracer -> MMFxTracer\r\n\r\n* fix bug when test with torch<1.12\r\n\r\n* update print log\r\n\r\n* fix lint\r\n\r\n* rm unuseful code\r\n\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: jacky <jacky@xx.com>\r\nCo-authored-by: Your Name <you@example.com>\r\nCo-authored-by: liukai <your_email@abc.example>\r\n\r\n* fix bug in placer holder (#395)\r\n\r\n* fix bug in placer holder\r\n\r\n* remove redundent comment\r\n\r\nCo-authored-by: liukai <your_email@abc.example>\r\n\r\n* Add get_prune_config and a demo config_pruning (#389)\r\n\r\n* update tools and test\r\n\r\n* add demo\r\n\r\n* disable test doc\r\n\r\n* add switch for test tools and test_doc\r\n\r\n* fix bug\r\n\r\n* update doc\r\n\r\n* update tools name\r\n\r\n* mv get_channel_units\r\n\r\nCo-authored-by: liukai <your_email@abc.example>\r\n\r\n* [Improvement] Adapt OFA series with SearchableMobileNetV3 (#385)\r\n\r\n* fix mutable bug in AttentiveMobileNetV3\r\n\r\n* remove unness code\r\n\r\n* update ATTENTIVE_SUBNET_A0-A6.yaml with optimized names\r\n\r\n* unify the sampling usage in sandwich_rule-based NAS\r\n\r\n* use alias to export subnet\r\n\r\n* update OFA configs\r\n\r\n* fix attr bug\r\n\r\n* fix comments\r\n\r\n* update convert_supernet2subnet.py\r\n\r\n* correct the way to dump DerivedMutable\r\n\r\n* fix convert index bug\r\n\r\n* update OFA configs & models\r\n\r\n* fix dynamic2static\r\n\r\n* generalize convert_ofa_ckpt.py\r\n\r\n* update input_resizer\r\n\r\n* update README.md\r\n\r\n* fix ut\r\n\r\n* update export_fix_subnet\r\n\r\n* update _dynamic_to_static\r\n\r\n* update fix_subnet UT & minor fix bugs\r\n\r\n* fix ut\r\n\r\n* add new autoaug compared to attentivenas\r\n\r\n* clean\r\n\r\n* fix act\r\n\r\n* fix act_cfg\r\n\r\n* update fix_subnet\r\n\r\n* fix lint\r\n\r\n* add docstring\r\n\r\nCo-authored-by: gaoyang07 <1546308416@qq.com>\r\nCo-authored-by: aptsunny <aptsunny@tongji.edu.cn>\r\n\r\n* [Fix]Dcff Deploy Revision (#383)\r\n\r\n* dcff deploy revision\r\n\r\n* tempsave\r\n\r\n* update fix_subnet\r\n\r\n* update mutator load\r\n\r\n* export/load_fix_subnet revision for mutator\r\n\r\n* update fix_subnet with dev-1.x\r\n\r\n* update comments\r\n\r\n* update docs\r\n\r\n* update registry\r\n\r\n* [Fix] Fix commands in README to adapt branch 1.x (#400)\r\n\r\n* update commands in README for 1.x\r\n\r\n* fix commands\r\n\r\nCo-authored-by: gaoyang07 <1546308416@qq.com>\r\n\r\n* Set requires_grad to False if the teacher is not trainable (#398)\r\n\r\n* add choice and mask of units to checkpoint (#397)\r\n\r\n* add choice and mask of units to checkpoint\r\n\r\n* update\r\n\r\n* fix bug\r\n\r\n* remove device operation\r\n\r\n* fix bug\r\n\r\n* fix circle ci error\r\n\r\n* fix error in numpy for circle ci\r\n\r\n* fix bug in requirements\r\n\r\n* restore\r\n\r\n* add a note\r\n\r\n* a new solution\r\n\r\n* save mutable_channel.mask as float for dist training\r\n\r\n* refine\r\n\r\n* mv meta file test\r\n\r\nCo-authored-by: liukai <your_email@abc.example>\r\nCo-authored-by: jacky <jacky@xx.com>\r\n\r\n* [Bug]Fix fpn teacher distill (#388)\r\n\r\nfix fpn distill\r\n\r\n* [CodeCamp #122] Support KD algorithm MGD for detection. (#377)\r\n\r\n* [Feature] Support KD algorithm MGD for detection.\r\n\r\n* use connector to beauty mgd.\r\n\r\n* fix typo, add unitest.\r\n\r\n* fix mgd loss unitest.\r\n\r\n* fix mgd connector unitest.\r\n\r\n* add model pth and log file.\r\n\r\n* add mAP.\r\n\r\n* update l1 config (#405)\r\n\r\n* add l1 config\r\n\r\n* update l1 config\r\n\r\nCo-authored-by: jacky <jacky@xx.com>\r\n\r\n* [Feature] Add greedy search for AutoSlim (#336)\r\n\r\n* WIP: add greedysearch\r\n\r\n* fix greedy search and add bn_training_mode to autoslim\r\n\r\n* fix cfg files\r\n\r\n* fix autoslim configs\r\n\r\n* fix bugs when converting dynamic bn to static bn\r\n\r\n* change to test loop\r\n\r\n* refactor greedy search\r\n\r\n* rebase and fix greedysearch\r\n\r\n* fix lint\r\n\r\n* fix and delete useless codes\r\n\r\n* fix pytest\r\n\r\n* fix pytest and add bn_training_mode\r\n\r\n* fix lint\r\n\r\n* add reference to AutoSlimGreedySearchLoop's docstring\r\n\r\n* sort candidate_choices\r\n\r\n* fix save subnet\r\n\r\n* delete useless codes in channel container\r\n\r\n* change files' name: convert greedy_search_loop to autoslim_greedy_search_loop\r\n\r\n* [Fix] Fix metafile (#422)\r\n\r\n* fix ckpt path in metafile and readme\r\n\r\n* fix darts file path\r\n\r\n* fix docstring in ConfigurableDistiller\r\n\r\n* fix darts\r\n\r\n* fix error\r\n\r\n* add darts of mmrazor version\r\n\r\n* delete py36\r\n\r\nCo-authored-by: liukai <your_email@abc.example>\r\n\r\n* update bignas cfg (#412)\r\n\r\n* check attentivenas training\r\n\r\n* update ckpt link\r\n\r\n* update supernet log\r\n\r\nCo-authored-by: aptsunny <aptsunny@tongji.edu.cn>\r\n\r\n* Bump version to 1.0.0rc2 (#423)\r\n\r\nbump version to 1.0.0rc2\r\n\r\nCo-authored-by: liukai <your_email@abc.example>\r\n\r\n* fix lint\r\n\r\n* fix ci\r\n\r\n* add tmp docstring for passed ci\r\n\r\n* add tmp docstring for passed ci\r\n\r\n* fix ci\r\n\r\n* add get_placeholder for quant\r\n\r\n* add skip for unittest\r\n\r\n* fix package placeholder bug\r\n\r\n* add version judgement in __init__\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\n* update prev commit\r\n\r\nCo-authored-by: LKJacky <108643365+LKJacky@users.noreply.github.com>\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: Yang Gao <Gary1546308416AL@gmail.com>\r\nCo-authored-by: kitecats <90194592+kitecats@users.noreply.github.com>\r\nCo-authored-by: Sheffield <49406546+SheffieldCao@users.noreply.github.com>\r\nCo-authored-by: whcao <41630003+HIT-cwh@users.noreply.github.com>\r\nCo-authored-by: jacky <jacky@xx.com>\r\nCo-authored-by: pppppM <67539920+pppppM@users.noreply.github.com>\r\nCo-authored-by: Yue Sun <aptsunny@tongji.edu.cn>\r\nCo-authored-by: zengyi <31244134+spynccat@users.noreply.github.com>\r\nCo-authored-by: zengyi.vendor <zengyi.vendor@sensetime.com>\r\nCo-authored-by: zhongyu zhang <43191879+wilxy@users.noreply.github.com>\r\nCo-authored-by: zhangzhongyu <zhangzhongyu@pjlab.org.cn>\r\nCo-authored-by: Xianpan Zhou <32625100+TinyTigerPan@users.noreply.github.com>\r\nCo-authored-by: Xianpan Zhou <32625100+PanDaMeow@users.noreply.github.com>\r\nCo-authored-by: Your Name <you@example.com>\r\nCo-authored-by: P.Huang <37200926+FreakieHuang@users.noreply.github.com>\r\nCo-authored-by: qiufeng <44188071+wutongshenqiu@users.noreply.github.com>\r\nCo-authored-by: wangshiguang <wangshiguang@sensetime.com>\r\nCo-authored-by: gaoyang07 <1546308416@qq.com>\r\nCo-authored-by: sunyue1 <sunyue1@sensetime.com>\r\nCo-authored-by: liukai <your_email@abc.example>\r\nCo-authored-by: Ming-Hsuan-Tu <qrnnis2623891@gmail.com>\r\nCo-authored-by: Yivona <120088893+yivona08@users.noreply.github.com>\r\nCo-authored-by: Yue Sun <aptsunny@alumni.tongji.edu.cn>\r\n\r\n* [Docs] Add docstring and unittest about backendconfig & observer & fakequant (#428)\r\n\r\n* add ut about backendconfig\r\n\r\n* add ut about observers and fakequants in torch\r\n\r\n* fix torch1.13 ci\r\n\r\n* [Docs] Add docstring for `MMArchitectureQuant` & `NativeQuantizer` (#425)\r\n\r\n* add docstring on mm_architecture& native_quantizer\r\n\r\n* add naive openvino r18 qat config & dist_ptq.sh\r\n\r\n* Added a more accurate description\r\n\r\n* unitest&doc\r\n\r\n* checkpoint url\r\n\r\n* unitest\r\n\r\n* passed_pre_commit\r\n\r\n* unitest on native_quantizer& fix bugs\r\n\r\n* remove dist_ptq\r\n\r\n* add get_placeholder&skipTest\r\n\r\n* complete arg descriptions\r\n\r\n* fix import bugs\r\n\r\n* fix pre-commit\r\n\r\n* add get_placeholder\r\n\r\n* add typehint and doctring\r\n\r\n* update docstring&typehint\r\n\r\n* update docstring\r\n\r\n* pre-commit\r\n\r\n* fix some problems\r\n\r\n* fix bug\r\n\r\n* [Docs] Add docstring and unitest about custom tracer (#427)\r\n\r\n* rename QConfigHandler and QSchemeHandler\r\n\r\n* add docstring about custom tracer\r\n\r\n* add ut about custom tracer\r\n\r\n* fix torch1.13 ci\r\n\r\n* fix lint\r\n\r\n* fix ci\r\n\r\n* fix ci\r\n\r\n* [Docs & Refactor] Add docstring and UT of other quantizers (#439)\r\n\r\n* add quantizer docstring and refactor the interface of AcademicQuantizer\r\n\r\n* add AcademicQuantizer unittest\r\n\r\n* add TensorRTQuantizer and OpenVINOQuantizer unittest & refactor prepare interface\r\n\r\n* adapt torch113 ci\r\n\r\n* fix import\r\n\r\n* fix lint\r\n\r\n* update some docstring\r\n\r\n* fix ci\r\n\r\n* [Feature&Doc]Modify ptq pipeline and support lsq (#435)\r\n\r\n* modify ptq pipeline and support lsq\r\n\r\n* use placeholder\r\n\r\n* fix lsq && quantloop\r\n\r\n* add lsq pytest\r\n\r\n* add quant loop pytest\r\n\r\n* test lsq observer\r\n\r\n* fix bug under pt13\r\n\r\n* fix reset_min_max_vals\r\n\r\n* fix bugs under pt13\r\n\r\n* fix configs\r\n\r\n* add get_qconfig_mapping\r\n\r\n* delete is_qat, add doc and fix pytest\r\n\r\n* delete useless codes in custom_tracer\r\n\r\n* skip pytest under pt13\r\n\r\n* add todo: check freezebn\r\n\r\n* fix pytest bugs\r\n\r\n* fix pytest\r\n\r\n* fix pytest\r\n\r\n* fix pytest\r\n\r\n* [Docs] Add customize_quantization_tutorial (#440)\r\n\r\n* [Docs] Add quantization user guide (#441)\r\n\r\n* add quantization user guide\r\n\r\n* fix layout\r\n\r\n* fix layout\r\n\r\n* update README\r\n\r\n* [Bug] Fix del redundant fakequant (#447)\r\n\r\nfix del redundant fakequant\r\n\r\n* [Feature] Add onnx exporters (#475)\r\n\r\n* fix del redundant fakequant\r\n\r\n* add onnx exporters\r\n\r\n* fix onnx exporters and add docstring\r\n\r\n* fix comments\r\n\r\n* delete useless codes\r\n\r\n* fix export_onnx in native quantizer\r\n\r\n---------\r\n\r\nCo-authored-by: pppppM <gjf_mail@126.com>\r\n\r\n* [Feature]Rewrite the origin model during prepare (#488)\r\n\r\n* add rewriter\r\n\r\n* add deploy_cfg arg\r\n\r\n* modify post_process_for_mmdeploy\r\n\r\n* fix bugs\r\n\r\n* add det config\r\n\r\n* [Feature] Using rewriter in mmrazor when building qmodels. (#490)\r\n\r\n* add rewriter\r\n\r\n* add deploy_cfg arg\r\n\r\n* modify post_process_for_mmdeploy\r\n\r\n* fix bugs\r\n\r\n* add det config\r\n\r\n* replace deepcopy\r\n\r\n* pop detectors' forward\r\n\r\n* [Feature] Quantization global optimization (#491)\r\n\r\n* add trtquantizer\r\n\r\n* unify all fakequant before deploy\r\n\r\n* move to aide\r\n\r\n* add yolox config\r\n\r\n* pre-rebase\r\n\r\n* add unittest\r\n\r\n* add a arg of post_process_for_deploy\r\n\r\n* test trt yolox deploy\r\n\r\n* opt quantizer interface\r\n\r\n* fix rebase\r\n\r\n* add trt r50 config\r\n\r\n* update trt setting\r\n\r\n* del redundant code\r\n\r\n* fix lint\r\n\r\n* fix ut of quantizers\r\n\r\n* del redundant file\r\n\r\n* fix lint\r\n\r\n* fix some comments\r\n\r\n* Fix code syntax in UT (#470)\r\n\r\nCo-authored-by: 王盟 <unicorn@MacBook-Pro.local>\r\n\r\n* passed lint and pytest\r\n\r\n* try to fix ci\r\n\r\n* [Bug] Try to fix CI (#502)\r\n\r\nfix lint\r\n\r\n* [Feature] Support lsq (#501)\r\n\r\n* support deploy_cfg=None\r\n\r\n* replace fakequant before load ckpt\r\n\r\n* add _load_from_state_dict to lsq fakequant\r\n\r\n* fix pre-commit\r\n\r\n* test lsq load state dict\r\n\r\n* change github ci: ubuntu 18.04 to ubuntu 20.04\r\n\r\n* get_deploy_model order change back\r\n\r\n* sync before save ckpt\r\n\r\n* delete strict=False\r\n\r\n* test context rewriter\r\n\r\n* fix pre commit config\r\n\r\n* try to fix ci\r\n\r\n* [Bug] Try to fix CI (#502)\r\n\r\nfix lint\r\n\r\n---------\r\n\r\nCo-authored-by: humu789 <humu@pjlab.org.cn>\r\nCo-authored-by: humu789 <88702197+humu789@users.noreply.github.com>\r\n\r\n* [Feature] Add exporter pytest (#504)\r\n\r\n* add exporter pytest\r\n\r\n* fix bugs\r\n\r\n* delete useless codes\r\n\r\n* handle onnx\r\n\r\n* delete useless codes\r\n\r\n* [Bug] Fix ci converage setting (#508)\r\n\r\nfix ci converage\r\n\r\n* [Bug] Fix codecov (#509)\r\n\r\n* remove codecov in requirements\r\n\r\n* try to fix ci\r\n\r\n* del adaround loss\r\n\r\n* [BUG] Fix quantization loop (#507)\r\n\r\n* fix quantization loop\r\n\r\n* fix quant loop\r\n\r\n* fix quant loop\r\n\r\n* fix qat configs\r\n\r\n* [Bug] Fix ci converage setting (#508)\r\n\r\nfix ci converage\r\n\r\n* [Bug] Fix codecov (#509)\r\n\r\n* remove codecov in requirements\r\n\r\n* try to fix ci\r\n\r\n* del adaround loss\r\n\r\n* add freeze_bn_begin to lsq\r\n\r\n* delete useless codes\r\n\r\n---------\r\n\r\nCo-authored-by: humu789 <88702197+humu789@users.noreply.github.com>\r\n\r\n* add test ptq\r\n\r\n* opt ptq pipeline\r\n\r\n* refactor quant configs\r\n\r\n* update config path\r\n\r\n* add summary analyse tool\r\n\r\n* fix benchmark_test:detnas_frcnn_shufflenet_subnet_coco_1x.py\r\n\r\n* update quantization README.md\r\n\r\n* update quantization metafile, readme, config path\r\n\r\n* update quantization docs\r\n\r\n* update git main link in workflow\r\n\r\n* update benchmark_summary_analyse.py\r\n\r\n* del dmcp results\r\n\r\n* [Bug] fix a rebase error (#514)\r\n\r\nfix a rebase error\r\n\r\n* [Bug] Fix CI (#515)\r\n\r\n* fix ci\r\n\r\n* mmcv2.0 need torch1.8+\r\n\r\n* Update CI config and Passed (#516)\r\n\r\n* test ci\r\n\r\n* update test.yml based on mmcv2.0.0\r\n\r\n* [Docs] Fix cwd test accuary (#517)\r\n\r\n* test ci\r\n\r\n* update test.yml based on mmcv2.0.0\r\n\r\n* update cwd_logits_pspnet result\r\n\r\n---------\r\n\r\nCo-authored-by: P.Huang <37200926+FreakieHuang@users.noreply.github.com>\r\nCo-authored-by: huangpengsheng <huangpengsheng@sensetime.com>\r\nCo-authored-by: LKJacky <108643365+LKJacky@users.noreply.github.com>\r\nCo-authored-by: liukai <liukai@pjlab.org.cn>\r\nCo-authored-by: Yang Gao <Gary1546308416AL@gmail.com>\r\nCo-authored-by: kitecats <90194592+kitecats@users.noreply.github.com>\r\nCo-authored-by: Sheffield <49406546+SheffieldCao@users.noreply.github.com>\r\nCo-authored-by: whcao <41630003+HIT-cwh@users.noreply.github.com>\r\nCo-authored-by: jacky <jacky@xx.com>\r\nCo-authored-by: pppppM <67539920+pppppM@users.noreply.github.com>\r\nCo-authored-by: FreakieHuang <frank0huang@foxmail.com>\r\nCo-authored-by: pppppM <gjf_mail@126.com>\r\nCo-authored-by: L-Icarus <30308843+L-Icarus@users.noreply.github.com>\r\nCo-authored-by: HIT-cwh <2892770585@qq.com>\r\nCo-authored-by: Yue Sun <aptsunny@tongji.edu.cn>\r\nCo-authored-by: zengyi <31244134+spynccat@users.noreply.github.com>\r\nCo-authored-by: zengyi.vendor <zengyi.vendor@sensetime.com>\r\nCo-authored-by: zhongyu zhang <43191879+wilxy@users.noreply.github.com>\r\nCo-authored-by: zhangzhongyu <zhangzhongyu@pjlab.org.cn>\r\nCo-authored-by: Xianpan Zhou <32625100+TinyTigerPan@users.noreply.github.com>\r\nCo-authored-by: Xianpan Zhou <32625100+PanDaMeow@users.noreply.github.com>\r\nCo-authored-by: Your Name <you@example.com>\r\nCo-authored-by: qiufeng <44188071+wutongshenqiu@users.noreply.github.com>\r\nCo-authored-by: wangshiguang <wangshiguang@sensetime.com>\r\nCo-authored-by: gaoyang07 <1546308416@qq.com>\r\nCo-authored-by: sunyue1 <sunyue1@sensetime.com>\r\nCo-authored-by: liukai <your_email@abc.example>\r\nCo-authored-by: Ming-Hsuan-Tu <qrnnis2623891@gmail.com>\r\nCo-authored-by: Yivona <120088893+yivona08@users.noreply.github.com>\r\nCo-authored-by: Yue Sun <aptsunny@alumni.tongji.edu.cn>\r\nCo-authored-by: Ivan Zhang <51170394+415905716@users.noreply.github.com>\r\nCo-authored-by: wm901115nwpu <wmnwpu@gmail.com>\r\nCo-authored-by: 王盟 <unicorn@MacBook-Pro.local>","shortMessageHtmlLink":"[Feature] Add MMRazor quantization (<a class=\"issue-link js-issue-link\" data-error-text=\"Failed to load title\" data-id=\"1670909440\" data-permission-text=\"Title is private\" data-url=\"https://github.com/open-mmlab/mmrazor/issues/513\" data-hovercard-type=\"pull_request\" data-hovercard-url=\"/open-mmlab/mmrazor/pull/513/hovercard\" href=\"https://github.com/open-mmlab/mmrazor/pull/513\">open-mmlab#513</a>)"}},{"before":null,"after":"1b51625bc104d17dafa7c980b74df91d91151428","ref":"refs/heads/remove_quant","pushedAt":"2023-04-22T05:20:48.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"aptsunny","name":"Yue Sun","path":"/aptsunny","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/36404164?s=80&v=4"},"commit":{"message":"add gml doc","shortMessageHtmlLink":"add gml doc"}}],"hasNextPage":false,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAADHgSlsgA","startCursor":null,"endCursor":null}},"title":"Activity · aptsunny/mmrazor"}