-
Fri, Oct 30, 2026 2
-
Recovery experiments R1-R4 done, claims endpoint, recovery section + figure, intro/conclusion/abstract drafts, release docs
…
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Main benchmark complete (288 runs), RC0094 rerun, rubric sensitivity (A-005), R2 stale-input, results draft, induced gate-rejection capture
…
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-