Skip to content

[quantization] Remove variant features#618

Merged
mhs4670go merged 2 commits intoSamsung:mainfrom
mhs4670go:refa
Apr 10, 2026
Merged

[quantization] Remove variant features#618
mhs4670go merged 2 commits intoSamsung:mainfrom
mhs4670go:refa

Conversation

@mhs4670go
Copy link
Copy Markdown
Contributor

This commit removes variant features.

Related: #612
TICO-DCO-1.0-Signed-off-by: seongwoo mhs4670go@naver.com

This commit removes variant features.

TICO-DCO-1.0-Signed-off-by: seongwoo <mhs4670go@naver.com>
@mhs4670go mhs4670go requested review from a team and stamalakhov April 10, 2026 10:45
with SuppressWarning(UserWarning, ".*"):
cm = tico.convert(
qlayer,
qlayer.wrapped.as_export_module("prefill").eval(),
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Ahh. @mhs4670go
But what about attention_mask and pos_embeds?

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

They are internally generated. Isn't it okay not to give them?

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

But they will be statically generated for every layer, and 2048 seq_len it will require much disk space. So i'm not sure. Externally set parameters were done on purpose to save disk space. But anyway it can be done later.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Ah, I forgot that. Let's revert this change.

Copy link
Copy Markdown
Contributor

@stamalakhov stamalakhov left a comment

Choose a reason for hiding this comment

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

LGTM! Thank you!

@mhs4670go mhs4670go merged commit 9a51755 into Samsung:main Apr 10, 2026
5 of 7 checks passed
@mhs4670go mhs4670go deleted the refa branch April 10, 2026 14:33
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.

2 participants