Skip to content

[Spec] Rename num_tokens_per_bs to num_tokens_per_req - #30977

Merged
hnyls2002 merged 6 commits into
mainfrom
lsyin/num-tokens-per-req
Jul 13, 2026
Merged

hnyls2002 merged 6 commits into
mainfrom
lsyin/num-tokens-per-req

Conversation

@hnyls2002

@hnyls2002 hnyls2002 commented Jul 13, 2026 •

Copy link
Copy Markdown
Collaborator

Mechanical rename (227 occurrences, 36 files): the value is the token count per request slot, not per batch — every use multiplies it by the request count, and EagleVerifyInput.num_tokens_per_req already uses the correct name for the same quantity. No behavior change; verified byte-identical to HEAD + textual substitution (plus one black line-wrap).


CI States

Latest PR Test (Base): 🚫 Run #29234563363
Latest PR Test (Extra): ❌ Run #29234563183

@gemini-code-assist

Copy link
Copy Markdown
Contributor

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

@hnyls2002

Copy link
Copy Markdown
Collaborator Author

/tag-and-rerun-ci

@github-actions github-actions Bot added the run-ci CI: run the baseline test suite on this PR label Jul 13, 2026
@hnyls2002
hnyls2002 requested a review from wisclmy0611 as a code owner July 13, 2026 08:08
@github-actions github-actions Bot added the documentation Improvements or additions to documentation label Jul 13, 2026
@mintlify

mintlify Bot commented Jul 13, 2026 •

Copy link
Copy Markdown
Contributor

Preview deployment for your docs. Learn more about Mintlify Previews.

Project Status Preview Updated (UTC)
lmsysorg 🟢 Ready View Preview Jul 13, 2026, 8:36 AM

💡 Tip: Enable Workflows to automatically generate PRs for you.

@hnyls2002
hnyls2002 merged commit c0f1f7e into main Jul 13, 2026
126 of 161 checks passed
@hnyls2002
hnyls2002 deleted the lsyin/num-tokens-per-req branch July 13, 2026 18:47
AliceChenyy added a commit to AliceChenyy/sglang that referenced this pull request Jul 14, 2026
Conflicts resolved:
- decode_cuda_graph_runner.py: kept the PP-proxy input-buffer refresh,
  adopted upstream's is_dflash_family()/is_ragged condition (sgl-project#30261),
  and renamed num_tokens_per_bs -> num_tokens_per_req in the PP proxy
  output slice (sgl-project#30977).
- eagle_worker_v2.py: dropped the local _get_plan_stream (extracted to
  a shared get_plan_stream upstream, sgl-project#31008/sgl-project#30857) and re-applied the
  pp_proxy_tensors parameter on the deduplicated
  forward_batch_generation signature.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Chronostasys pushed a commit to MindLab-Research/sglang that referenced this pull request Aug 24, 2026

This branch was successfully deployed

1 active deployment
staging - docs_new — 66c6d939 Deployed Jul 13, 2026 by mintlify[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

blackwell SM100/SM120 deepseek documentation Improvements or additions to documentation lora npu run-ci CI: run the baseline test suite on this PR speculative-decoding

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant