Skip to content

[MPQEditor] Implement 'dispose' and 'onViewMessage' #1539

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

Merged
merged 1 commit into from
Apr 28, 2023

Conversation

stamalakhov
Copy link
Contributor

This commit implements 'dispose' and 'onViewMessage' for manual mixed precision quantization task.

Its correctness is tested in #1530

Fresh draft: #1530
Full draft: #1505
Related: #1491

ONE-vscode-DCO-1.0-Signed-off-by: s.malakhov [email protected]

This commit implements 'dispose' and 'onViewMessage' for manual mixed precision quantization task.

ONE-vscode-DCO-1.0-Signed-off-by: s.malakhov <[email protected]>
@stamalakhov stamalakhov self-assigned this Apr 27, 2023
// TODO
public onViewMessage(message: any) {
switch (message.command) {
case MessageDefs.selection:
Copy link
Contributor

@dayo09 dayo09 Apr 28, 2023

Choose a reason for hiding this comment

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

I've got some UI feedbacks. If it's not easy to handle on this pr, please may you consider it later.

image

On #1505 draft,

  • when I click on 'add' node, MPQEditor seems to adds the node to the mpq editor slot.
  • when I click on 'Place holder's or 'Add 2' node, as they are not the mpq target, they shows the property view on the right side of the circle editor.

I think that 'click' event must result in the same behavior.
We can choose either to open the property view same to 'add' operation or not to open the property view to other operations.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

@dayo09
Thank you. I'll try to fix it in the draft at first.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

@dayo09 @seanshpark
The same behaviour in the netron:
click on input/output nodes shows Model Properties, while clicking on the internal nodes show Nodes Properties.
For MPQ draft clicking internal nodes results in their selection for editing, while for input/output nodes still shows Model Properties to observe input/output size and theirs type.
To handle this issue:

  1. We can hide input/output nodes
  2. We can disable clicking input/output nodes

But still there may be some issues that some users are used to observe input/output size.
IMHO.

Copy link
Contributor

Choose a reason for hiding this comment

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

Or 3. We can show Properties too, when clicking on the internal nodes(ex. Add).

Copy link
Contributor

Choose a reason for hiding this comment

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

I will approve this PR, let's go on about it on the other issue or PR. Thanks for listening to me.!

Copy link
Contributor Author

Choose a reason for hiding this comment

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

I will approve this PR, let's go on about it on the other issue or PR. Thanks for listening to me.!

Thank you very much!

@dayo09
Copy link
Contributor

dayo09 commented Apr 28, 2023

Hi, this is another UI feedback.
It's not related to this PR, but rather related to the draft #1505.

image

It seems that : In the (1) area, it decides the default quantization configuration and (2) area specifies the per-node(layer) quantization configuration.

At first look, I didn't understand what the (1) area would do. From my point of view, I propose to add 'Set default quantization' 'Set specific layer's quantization' in each areas.

@stamalakhov
Copy link
Contributor Author

Hi, this is another UI feedback. It's not related to this PR, but rather related to the draft #1505.

image

It seems that : In the (1) area, it decides the default quantization configuration and (2) area specifies the per-node(layer) quantization configuration.

At first look, I didn't understand what the (1) area would do. From my point of view, I propose to add 'Set default quantization' 'Set specific layer's quantization' in each areas.

@dayo09
Thank you. I'll try to fix it in the draft at first.

@stamalakhov
Copy link
Contributor Author

It seems that : In the (1) area, it decides the default quantization configuration and (2) area specifies the per-node(layer) quantization configuration.

@dayo09
To fix layout it needs to edit media files, so may be we can handle it in another PR?

@stamalakhov stamalakhov requested a review from dayo09 April 28, 2023 05:01
@dayo09
Copy link
Contributor

dayo09 commented Apr 28, 2023

@stamalakhov

To fix layout it needs to edit media files, so may be we can handle it in another PR?

Sure :-D Thanks a lot :-D

Copy link
Contributor

@seanshpark seanshpark left a comment

Choose a reason for hiding this comment

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

LGTM

@seanshpark seanshpark merged commit 40dd14b into Samsung:main Apr 28, 2023
@stamalakhov stamalakhov deleted the manual_MPQ_the_rest_br branch April 28, 2023 07:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants