Repository navigation
Evidence case: self-hosted OpenAI-compatible server, client vs server surface (Discussion #26) - #39
Open
qaisermehdi3-coder wants to merge 1 commit into
Open
qaisermehdi3-coder wants to merge 1 commit into
qaisermehdi3-coder wants to merge 1 commit into
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What & why
Adds a runnable evidence case for known-limits section 6, as offered in Discussion #26. It pins one live capture (vLLM 0.27.1 on a T4): the same request three times, one of them a stream without stream_options. Client surface: 70 / 32 on two calls, one call with no usage. Server surface: 3 requests, 105 / 48, and 64 prompt tokens read from the prefix cache. Gap: +35 / +16.
It also pins a second client-surface gap: vLLM sends prompt_tokens_details only with --enable-prompt-tokens-details (off by default), so the cache split is unreported, not 0. And one small refinement: with exactly one absent call the gap is that call's by elimination; with two or more it stays UNPROVABLE.
Evidence
python3 conformance/evidence/selfhosted-openai-compatible-reconcile/run_case.py
Expected: 12 PASS lines, then CONFORMANT. The raw files (full /metrics snapshots, the client's usage per call, the server log) are published unedited at qaisermehdi3-coder/qvunex@1094fd5, with sha256 in expected.json.
Conformance impact
New evidence case only (4 files). No spec, registry or adapter change. CI: a one-step addition to conformance.yml after the unknown-model-pricing case would run it; I left the workflow untouched.
AI-assisted: yes
The runner, the fixture extraction and the README were written with an AI assistant. I did the run live. Every fixture value was checked by script against the raw files, and each check in the runner fails when the rule it guards is removed (10 of 10).