✓ No known CVEs patched in this version
Topics
+10 more
Summary
AI summaryFixed evals bugs: correct per‑class weighted F‑score, count executor timeouts toward retry limit, and conditionally enable 0/1 label detection.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Bugfix | Medium |
Counts AsyncExecutor timeouts against max_retries in evals. Counts AsyncExecutor timeouts against max_retries in evals. Source: llm_adapter@2026-07-14 Confidence: high |
— |
| Bugfix | Medium |
Gates 0/1 positive_label auto-detection on default macro average in evals. Gates 0/1 positive_label auto-detection on default macro average in evals. Source: llm_adapter@2026-07-14 Confidence: high |
— |
| Bugfix | Medium |
Computes macro/weighted F-score per class to match sklearn semantics in evals. Computes macro/weighted F-score per class to match sklearn semantics in evals. Source: llm_adapter@2026-07-14 Confidence: high |
— |
Full changelog
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
Related context
Related tools
Earlier breaking changes
- varize-phoenix-v19.0.0 GraphQL createUserApiKey and createSystemApiKey reject API-key callers.
- varize-phoenix-v18.0.0 Changes session time-range filters to use interval-overlap semantics.
- varize-phoenix-v17.0.0 Adds system settings for admin-managed assistant enablement and trace recording policy
- varize-phoenix-v16.0.0 Sandboxing and Code Evaluators introduce breaking changes in Phoenix v16.0.0.
- varize-phoenix-v15.7.0 Removes v1 /chat route and associated code
Beta — feedback welcome: [email protected]