Issues

• Option length predicts the correct answer in PersonaMem-v2: gold is among the shortest options in all 52 sampled who=others rows and among the longest in 46.7% overall

  bowen-upenn/PersonaMem#42 · Filter by author Paul-Kyle opened on Aug 8, 2026

• A no-context baseline scores 49.2% on PersonaMem-128k (v1), including 80% on track_full_preference_evolution

  bowen-upenn/PersonaMem#41 · Filter by author Paul-Kyle opened on Aug 6, 2026

• Why dont you give the conversation time？

  bowen-upenn/PersonaMem#37 · Filter by author RichardWang11 opened on Mar 5, 2026

• Reproducing PersonaMem-v2 Text Benchmark Results

  bowen-upenn/PersonaMem#36 · Filter by author yihangyao opened on Feb 26, 2026

• Unused shared contexts in PersonaMem-128k

  bowen-upenn/PersonaMem#34 · Filter by author xingerazi opened on Jan 4, 2026

• About PersonaMem-V2 code

  bowen-upenn/PersonaMem#33 · Filter by author optimize-spatial-scale opened on Dec 24, 2025

• About ImplicitPersona (PersonaMem V2)

  bowen-upenn/PersonaMem#32 · Filter by author IshiKura-a opened on Nov 28, 2025