Add DSpark algorithm page

[guan404ming]

Purpose

DSpark is supported in the trainer but has no algorithm page (followup from #840). Adds documentation covering the Markov and confidence heads, the DSpark-specific flags and their defaults, and wires it into the nav, the algorithms index, and the decision guide.

cc @WindChimeRan

Tests

Docs-only. make quality passes.

Checklist

I have filled in:

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan/results, such as providing test command and pasting the results.
• (Optional) The necessary documentation update.
• I (a human) have written or reviewed the code in this pr to the best of my ability.

[coderabbitai]

Important

Review skipped

Auto incremental reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: c6410384-5f6a-4641-83e0-aa36afbcb519

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

📝 Walkthrough

Walkthrough

Changes

DSpark documentation

Layer / File(s)	Summary
DSpark guide docs/user_guide/algorithms/dspark.md	Adds documentation for DSpark’s Markov and confidence heads, behavior, configuration, training, model availability, citation, and related guides.
Algorithm documentation integration docs/user_guide/algorithms/index.md, docs/user_guide/algorithms/decision_guide.md, docs/.nav.yml	Adds DSpark to the algorithm list, navigation, support matrix, recommendations, and detail links. Documents five supported speculative decoding algorithms.

Possibly related PRs

• vllm-project/speculators#840: Updates overlapping DSpark documentation surfaces and training documentation.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly and concisely matches the main change: adding a DSpark algorithm page.
Description check	✅ Passed	The description is directly related to the docs changes and accurately summarizes the DSpark page and navigation updates.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 4

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/user_guide/algorithms/dspark.md`:
- Around line 17-19: Update the “Confidence Head” section to explain that
confidence scores are conditional on preceding draft-token acceptance, that
DSpark cumulatively computes prefix-survival probabilities, and that a
hardware-aware scheduler selects verification length rather than applying
independent scores or a simple threshold. Include a concise example consistent
with the current API and speculative decoding behavior.
- Line 3: Update the DSpark architecture description around the vLLM deployment
statement to remove the claim that its deployment path is unchanged. State that
the workflow is similar to DFlash but requires DSpark-aware vLLM serving support
and configuration, including the dedicated dspark method and sequential Markov
sampling.
- Around line 37-39: Update the “Pretrained Models” section in dspark.md to
remove the claim that no pretrained DSpark models exist. Link to the available
RedHatAI DSpark checkpoints, including RedHatAI/GLM-5.2-speculator.dspark, or
explicitly limit the statement to models provided within this repository while
retaining the training example.
- Line 15: Update the DSpark algorithm documentation around the `--markov-rank
0` guidance to include `--no-confidence-head-with-markov`, and also document
`--no-enable-confidence-head` when pure DFlash behavior is intended. Ensure the
documented command-line options produce a valid `DSparkDraftModel` configuration
with `markov_rank=0`.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: d371da7f-b138-4cbc-93df-200be22e80e7

📥 Commits

Reviewing files that changed from the base of the PR and between a9bc119 and 7a86bf3.

📒 Files selected for processing (4)

• docs/.nav.yml
• docs/user_guide/algorithms/decision_guide.md
• docs/user_guide/algorithms/dspark.md
• docs/user_guide/algorithms/index.md

[guan404ming]

Pointed the GLM link at the actual repo (GLM-5.2-speculator.dspark-preview) and dropped the gemma row since RedHatAI/gemma-4-31B-it-speculator.dspark doesn't exist yet. Happy to add it back once it lands under RedHatAI.

[mergify]

Merge Protections

🟢 Merge protection satisfied — ready to merge.

Show 1 satisfied protection

🟢 Require approval from approved reviewers list

All pull requests must have at least one approving review from a member of the approved reviewers list before merging.

• any of:
  • approved-reviews-by = fynnsu
  • approved-reviews-by = dsikka
  • approved-reviews-by = orestis-z
  • approved-reviews-by = rahul-tuli
  • approved-reviews-by = shanjiaz

[mergify]

The quality checks have failed. Please run make style and make quality under
the root directory to address the lint failures. You will need to install the
dev optional install to get the required linting packages:
https://github.com/vllm-project/speculators/blob/main/CONTRIBUTING.md

[guan404ming]

Fixed the quality issue. Please take another look, thanks!