-
Notifications
You must be signed in to change notification settings - Fork 446
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.
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
move mmdet templates #985
move mmdet templates #985
Conversation
run ote_sdk tests |
ote-test/1451/ |
ote-test/1452/ |
run ote_sdk tests |
ote-test/1453/ |
run ote-test |
@@ -31,6 +31,7 @@ def parse_args(): | |||
"--root", help="A root dir where templates should be searched.", default="." | |||
) | |||
parser.add_argument("--task_type") | |||
parser.add_argument("--experimental", action="store_true") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe turn it on by default?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
maybe, but in this case I would set it to False while checking ote_cli features like train, eval, export and etc. Otherwise testing will take much more longer. However experimental templates have to be tested as well, I don't know how it can be done during reasonable time.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is testing code, this is a user-facing API. And IMO having this feature disabled here means that we don't want users to use "experimental" templates in OTE. So, I think about switching the default.
This being said, I agree that non-experimental templates should be tested in the first place, and experimental ones may be skipped to decrease CI load.
@@ -0,0 +1,42 @@ | |||
# Copyright (C) 2021 Intel Corporation | |||
# |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just as a discussion point.
- As we now have common set of tasks/utils/whatever for all mmdetection-powered templates, it looks safe to remove extra level of nesting (
detection
) here. But probably not in this PR. - Having
mmdetection
directory containingdetection_tasks
makes the story a bit complicated. Maybe we should not reference any particular algo backend in the folders' tree of OTE at all and leave it being hidden under the neutralsubmodule
? I foresee difficulties if we end up having multiple detection backends though. Any thoughts on this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
- makes sense
- not sure that I fully understand, but "external" and "mmdetection" (and other) names are kept here because of CI. I made this transition so that all paths such as "external/mmdetection/init_venv.sh", "external/mmdetection/tests", etc are not changed. I guess these paths are hardcoded, @LeonidBeynenson , right? Or are you, @druzhkov-paul , telling about different things?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You captured my point right.
I see, that we have limitation at the CI side. @LeonidBeynenson could you please estimate how difficult and time-consuming would that be to support modified directories structure? I mean 100% transition, not "almost completed" state.
At the same time do we have an agreement that such renaming is a reasonable thing to do? Or it does not pay off taking into account all of the required changes in the infrastructure?
run ote_sdk tests |
run ote_sdk tests |
ote-test/1468/ 🟡 |
No description provided.