Skip to content

refactor: centralize logits processing in causal models. - #2613

Draft
wai-bao wants to merge 1 commit into
xLLM-AI:mainfrom
wai-bao:refactor/hoist-draft-weight-sharing-v2
Draft

wai-bao wants to merge 1 commit into
xLLM-AI:mainfrom
wai-bao:refactor/hoist-draft-weight-sharing-v2

Conversation

@wai-bao

@wai-bao wai-bao commented Oct 29, 2026 •

Copy link
Copy Markdown
Collaborator

LLM and VLM workers use the model-owned LogitsProcessor to select hidden rows and project them through the existing LM head. Python models use the same runtime path and supply their own LM head. The processor can return the selected rows alongside logits for speculative decoding.

Remove the old model-level logits methods and Python compute_logits overrides.

@wai-bao
wai-bao force-pushed the refactor/hoist-draft-weight-sharing-v2 branch from 3fe5d5d to a1a250a Compare October 29, 2026 13:17
@netlify

netlify Bot commented Oct 29, 2026 •

Copy link
Copy Markdown

✅ Netlify Deploy Preview ready

Built without sensitive environment variables

Name Link
🔨 Latest commit 81a7172
🔍 Latest deploy log https://app.netlify.com/projects/xllm-docs-preview/deploys/6b027de4b76c5e87ebf09494
😎 Deploy Preview https://deploy-preview-2613--xllm-docs-preview.netlify.app
📱 Preview on mobile
Toggle QR Code...

QR Code

Use your smartphone camera to open QR code link.
🤖 Make changes Run an agent on this branch

To edit notification comments on pull requests, go to your Netlify project configuration.

@wai-bao
wai-bao force-pushed the refactor/hoist-draft-weight-sharing-v2 branch 28 times, most recently from fd1860d to 4784751 Compare November 1, 2026 21:43
@wai-bao
wai-bao force-pushed the refactor/hoist-draft-weight-sharing-v2 branch 26 times, most recently from d097460 to 63d7bdc Compare November 7, 2026 13:14
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 18, 2026 •

Copy link
Copy Markdown
Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-18T08:03:32.363170Z ce0dbdb Draft marked ready
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ce0dbdbc39

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread xllm/models/llm/deepseek_v4_dspark.h Resolved
Comment thread xllm/core/framework/model/causal_lm.h Resolved
Comment thread xllm/core/runtime/executor_impl.h Resolved
Comment thread xllm/models/llm/mlu/deepseek_v4_mtp.h Resolved

This branch has not been deployed

No deployments
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