* add pytorch 1.1.0 SyncBN support

* change BatchNorm2d to _BatchNorm and call freeze after train

* add freeze back to init function

* fixed indentation typo in adding freeze

* use SyncBN protect member func to set ddp_gpu_num

* Update README.md

update pytorch version to 1.1

* support param-wise optimizer settings

* fix outdated docstring

* minor fix for param checking

* use_sigmoid_cls -> cls_sigmoid_loss; use_focal_loss -> cls_focal_loss

* fix flake8 error

* cls_sigmoid_loss - > use_sigmoid_cls

* rename normalize to norm_cfg

* update configs

* Update resnet.py

fix bug of show_result

* rewrite multigpu testing with mmddp

* fix an indent

* update slurm testing script

* update readme and scripts

* use different configs for proposal train/test

* update configs for dcn

* update inf speed

* update inf speed in dcn, gn, htc

* update comparision

* keep backward compatibility

change frozen to requires_grad for clear behavior

* update installation guide

* fix typo

* update conda python version from 3.6 to 3.6

Fix typo in README.md

Update readme and tutorials

Reorg model zoo and configs

* fix a bug in basicblock

* fix padding and dilation in basic block

* remove stride and dilation in conv2

* compared with maskrcnn-benchmark instead detectron.pytorch, delete some outdated sentences

* add memory data

* Update MODEL_ZOO.md

* delete some wrong sentences

* fix speed in comparison section

* minor fix

* new model zoo

* update model url and performance

* update pytorch version

fix typo in MODEL_ZOO on Mask R-CNN

* support GN in RetinaHead

* add conv_cfg argument for RetinaHead

* add model upgrading tool and update retinanet model urls

* minor fix for regex strings

Check dataset type by subclass instead of names

Bug fix for freezing BN parameters

Add a tool to process models to be published

* add support for weight standardization

* add ws support for htc heads

* add a config file for 20-23-24e lr schedule

* update readme of weight standardization

Fixed error `TypeError: super(type, obj): obj must be an instance or subtype of type` .

* add HybridTaskCascade

* add configs for other backbones

* add keep_ratio argument for segmap transform

* add readme for HTC

* fix linting errors

* split assign and sampling as in Cascade R-CNN

* remove unused imports

* add a large model

* update htc

* add two stage w/o neck and w/ upperneck

* add rpn r50 c4

* update c4 configs

* fix

* config update

* update config

* minor update

* mask rcnn support c4 train and test

* lr fix

* cascade support upper_neck

* add cascade c4 config

* update config

* update

* update res_layer to new interface

* refactoring

* c4 configs update

* refactoring

* update rpn_c4 config

* rename upper_neck as shared_head

* update

* update configs

* update

* update c4 configs

* update according to commits

* update

* implement _init_dist_slurm()

* add slurm train/test scripts

* fix linting error

* minor fix

Create a new release v0.6.0