Skip to content

Execute direct hybrid layers across PP and VPP - #6297

Draft
Phlip79 wants to merge 1 commit into
NVIDIA:pull-request/6295from
Phlip79:philip/direct-hybrid-pp-vpp
Draft

Execute direct hybrid layers across PP and VPP#6297
Phlip79 wants to merge 1 commit into
NVIDIA:pull-request/6295from
Phlip79:philip/direct-hybrid-pp-vpp

Conversation

@Phlip79

@Phlip79 Phlip79 commented Aug 5, 2026

Copy link
Copy Markdown
Member

Summary

  • Build direct specs with existing layer classes and per-occurrence configs.
  • Execute explicit PP/VPP chunks with global numbering, checkpoint keys, and compatible RoPE inputs.
  • Preserve legacy parser, selector, shared config, layer kwargs, and ownership behavior.

Stacked on #6295.

@copy-pr-bot

copy-pr-bot Bot commented Aug 5, 2026

Copy link
Copy Markdown

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

Phlip79 commented Aug 5, 2026

Copy link
Copy Markdown
Member Author

/ok to test d037034

@Phlip79

Phlip79 commented Aug 5, 2026

Copy link
Copy Markdown
Member Author

/ok to test a353a76

@Phlip79

Phlip79 commented Aug 5, 2026

Copy link
Copy Markdown
Member Author

/ok to test 89d2874

Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
@Phlip79
Phlip79 force-pushed the philip/direct-hybrid-pp-vpp branch from 89d2874 to e344ee2 Compare August 5, 2026 23:18
@Phlip79

Phlip79 commented Aug 5, 2026

Copy link
Copy Markdown
Member Author

/ok to test e344ee2

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.

1 participant