Repository navigation
Test the latency harness and fix a zero-rounded TTFT being dropped - #41
Merged
Merged
Conversation
The first streamed token can round to 0.0 s against a low-latency endpoint. `if first_any` then treats that measurement as "never arrived" and the run is reported with a null TTFT; the TPOT aggregation had the same truthiness test. Both now compare against None, so a fast first token survives. Add tests that exercise the harness without any accelerator: a scripted SSE server covers the two thinking-field spellings, a thinking-only stream that must leave the content clock unset, header-only bearer auth, per-request failure counting, and the fact that receipts carry no host address. Also ship tools/bench/config.example.json so a run can record the serving configuration next to its numbers instead of in a separate note.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Follow-up to #40.
if first_anytreated a first token that rounds to 0.0 s as missing, so a fast endpoint reported a null TTFT; the TPOT aggregation used the same truthiness test. Both now compare againstNone.tests/test_bench.pydrives the harness against a scripted SSE server, so it needs no accelerator: it covers both thinking-field spellings, a thinking-only stream leaving the content clock unset, header-only bearer auth, per-request failure counting, payload assembly for snippets/haystack/images/tools/response_format, receipt de-hosting, and that every shipped case loads with a unique id.tools/bench/config.example.jsongives a run a place to record the serving configuration next to its numbers. 243 tests pass.