docs: align serving docs with current APIs - #3366
Open
localai-org-maint-bot wants to merge 4 commits into
Open
localai-org-maint-bot wants to merge 4 commits into
localai-org-maint-bot wants to merge 4 commits into
Conversation
The server reference omits shipped routes and contradicts the HTTP response builders. Record a CPU-only documentation scope before implementation for ISSUE-LOCAL-01M3TPN0QZ8856GY1MHN2TKNHX. FOLLOWING_AGENTS_PROTOCOL Following-Agents-Protocol: true AI-Assisted: true Assisted-by: Codex:GPT-6 [Codex]
The guide still says HTTP cannot return prompt log probabilities. Describe the response fields, decision routes, and EOS fallback from current source. Link the Tev1 recipe from the news section so readers can find the new decision path and its verification limits. This is the public prose repair scoped by ISSUE-LOCAL-01M3TPN0QZ8856GY1MHN2TKNHX. No runtime or benchmark changes are included. README, site, and changed-link checks pass. The benchmark index retains the same 16 orphan-file failures as the untouched base. FOLLOWING_AGENTS_PROTOCOL Following-Agents-Protocol: true AI-Assisted: true Assisted-by: Codex:GPT-6 [Codex]
The duplicate EOS prose says an absent generation config adds nested IDs. The outer config takes precedence, so that claim can misstate the stop set. Link the authoritative reference while retaining primary and secondary IDs and the effect of ignore_eos in the usage guide. Tracks ISSUE-LOCAL-01M3TPN0QZ8856GY1MHN2TKNHX. The operator reruns full preflight for this documentation repair. This helper runs focused checks without repeating that expensive suite. FOLLOWING_AGENTS_PROTOCOL Following-Agents-Protocol: true AI-Assisted: true Assisted-by: Codex:GPT-6 [Codex]
Record the reviewed server-guide repair and its CPU-only checks for ISSUE-LOCAL-01M3TPN0QZ8856GY1MHN2TKNHX. Preserve the existing GPU and runtime-validation limits. The full preflight fails on existing repository and environment checks. This documentation-only handoff uses source, structure, link, and exact-base checks. It is not a green integration-readiness result or authorization to merge. FOLLOWING_AGENTS_PROTOCOL Following-Agents-Protocol: true AI-Assisted: true Assisted-by: Codex:GPT-6 [Codex]
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Row
Documentation repair tracked by
ISSUE-LOCAL-01M3TPN0QZ8856GY1MHN2TKNHXand.agents/specs/docs-server-surface-refresh.md. No model or engine lifecycle transition.Before starting
mainatfce36733bad6in a separate worktree.api_server.cpp, callback wiring inserver_main.cpp, completion response builders and validators, EOS configuration readers, and their existing tests.What changed
The server reference omitted shipped decision endpoints and incorrectly said HTTP responses omit prompt log probabilities. Document their registration conditions, response fields, and request limits. Link the usage page to the detailed reference. Explain EOS fallback precedence and add Tev1 news linking its setup and validation limits.
Evidence
scripts/agent-preflight.shpasses. It does not pass on the base; see Honest gaps.git diff --check.python3 -m unittest tests.scripts.test_check_readme_structure: 19 tests pass.python3 scripts/check-tree-compiles.py --base fce36733b: no C++ source, header, or build file in scope.Independent review checked the prose against code and existing tests. Scratch mutations of a link and a heading failed the scoped checker and restored byte-for-byte.
Speed claims
Honest gaps
The operator preflight exits 1 with 51 failed checks and 14 skips, including existing repository drift and missing Alpine prerequisites. Its range checks use stale
origin/main; style, trailer, and compilation-scope checks pass against the actual basefce36733b. The benchmark-index checker already reports 16 orphan detail files. This PR does not claim a green full preflight, new runtime parity, or GPU verification. Its documentation claims were checked against current source and existing tests.FOLLOWING_AGENTS_PROTOCOL
Following-Agents-Protocol: true
AI-Assisted: true
Assisted-by: Codex:GPT-6 [Codex]