Reproduce “Instant Personalized Large Language Model Adaptation via Hypernetwork” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhaoxuan Tan, Zixuan Zhang, Haoyang Wen, Zheng Li, Rongzhi Zhang, Pei Chen, Fengran Mo, Zheyuan Liu and 3 more
P2P generates personalized LoRA adapters from user profiles, replacing per-user adapter training at deployment. A hypernetwork trained across users encodes profile information and produces adapter parameters in one pass. On random LaMP/LongLaMP tasks, it averages 0.580 classification accuracy and 0.244 ROUGE-L, versus OPPU’s 0.568 and 0.221; on constructed OOD users, classification leads OPPU, but generation ROUGE-L trails Full History.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.