Skip to content

feat(classification): Update classification codebase to 1.0 #71

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 6 commits into from
Oct 19, 2020
Merged

feat(classification): Update classification codebase to 1.0 #71

merged 6 commits into from
Oct 19, 2020

Conversation

@zhouyizhuang-megvii
Copy link
Collaborator

@zhouyizhuang-megvii zhouyizhuang-megvii commented Oct 15, 2020

No description provided.

wjfwzzc
wjfwzzc reviewed Oct 15, 2020
View reviewed changes
@wjfwzzc
Copy link
Collaborator

wjfwzzc commented Oct 15, 2020

别忘把pylint修了。

@zhouyizhuang-megvii zhouyizhuang-megvii changed the base branch from master to imperative-runtime October 15, 2020 11:56
wjfwzzc
wjfwzzc requested changes Oct 19, 2020
View reviewed changes
Copy link
Collaborator

@wjfwzzc wjfwzzc left a comment •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

README没更新，尤其是很多参数已经变了。
另外代码似乎是多机全占的逻辑，明确不支持单机单/多卡了是么？

other LGTM.

@wjfwzzc wjfwzzc changed the title Update classification codebase to 1.0 feat(classification): Update classification codebase to 1.0 Oct 19, 2020
@zhouyizhuang-megvii
Copy link
Collaborator Author

zhouyizhuang-megvii commented Oct 19, 2020

另外代码似乎是多机全占的逻辑，明确不支持单机单/多卡了是么？

现在支持--ngpus参数了，README也更新好了

@wjfwzzc
Copy link
Collaborator

wjfwzzc commented Oct 19, 2020 •
edited
Loading

还差一点点，当world_size = 1 & ngpus = 1时，init_process_group是否有不work的可能？
如果不work要加特判的话，还有其它dist的环节可能都需要check，比如gm的callback，bcast_list_和all_reduce_sum。
建议真实跑起来看看吧……

还有，默认dist.make_allreduce_cb("MEAN")，不SUM了？

@zhouyizhuang-megvii
Copy link
Collaborator Author

zhouyizhuang-megvii commented Oct 19, 2020

还差一点点，当world_size = 1 & ngpus = 1时，init_process_group是否有不work的可能？
如果不work要加特判的话，还有其它dist的环节可能都需要check，比如gm的callback，bcast_list_和all_reduce_sum。
建议真实跑起来看看吧……

还有，默认dist.make_allreduce_cb("MEAN")，不SUM了？

单卡的场景可以正常跑了；也已修改为SUM

wjfwzzc
wjfwzzc approved these changes Oct 19, 2020
View reviewed changes
Copy link
Collaborator

@wjfwzzc wjfwzzc left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM.

@wjfwzzc wjfwzzc merged commit 4a93a77 into MegEngine:imperative-runtime Oct 19, 2020
wjfwzzc pushed a commit that referenced this pull request Oct 21, 2020
@zhouyizhuang-megvii @dc3671
feat(classification): Update classification codebase to 1.0 (#71)
134ad00 
* fix(quant): modify elemwise add to fuse_add_relu (#64)

* update classification codebase to 1.0

* Fix style according to reviews

* update README and command line args

* fix single gpu logic

* use all reduce sum mode

Co-authored-by: Dash Chen <chenzhh3671@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@wjfwzzc wjfwzzc wjfwzzc approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@zhouyizhuang-megvii @wjfwzzc @dc3671