Skip to content

v1.6.1 — token cap fix, quota message, reboot scripts

Choose a tag to compare

View all tags
@wilke wilke released this 15 Sep 12:13
· 120 commits to main since this release
v1.6.1
4ea2e38
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Patch release for the hackathon week, on top of v1.6.0.

  • Generation token cap no longer silently truncates tables (#553): prompt templates carry a per-template max_output_tokens (the v2 fixture sets 2500 on the three table templates); a cut-off templated /v1/query answer now carries truncated: true; LLM_MAX_OUTPUT_TOKENS is a real setting (default 512) that also applies to plain answers. An untemplated /v1/query still returns exactly {answer, sources, rewritten_queries}.
  • Collection quota message tells a user what works (#557, fixes #555): at the cap the UI no longer suggests renaming.
  • coconut reboot scripts know the hackathon tenant (#559): its stores (incl. the dedicated postgres), its API, and RAGSTACK_GIT_TAG/SHA on every API launch.

Upgrading from v1.6.0: check out the tag, restart the API, rebuild static UIs; tenants with PROMPT_TEMPLATES_FILE should replace their file with contracts/fixtures/prompt-templates.example.yaml from this tag (v2 templates). Revert point: v1.6.0.

Assets 2
Loading