Reproduce “Demonstration of MANA: Multi-perspective Alignment of NPC Agents” 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

Amir Tahmasbi, Keshav Singh, Shaheer Raza, Owen Mei, Jason Wu

MANA lets game developers teach NPC behavior by taking over a character and demonstrating actions and dialogue as situations arise during play. An LLM turns those demonstrations into a natural-language persona and an interpretable action policy, which guides autonomous behavior. In a Minecraft demonstration, two NPCs developed contrasting behaviors after roughly 2 minutes; this result is not a controlled evaluation.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.