Natural language boosts LLM performance in coding, planning,

MIT CSAIL researchers create three neurosymbolic methods to help language models build libraries of better abstractions within natural language: LILO assists with code synthesis, Ada helps with AI planning, and LGA guides robotic navigation and manipulation.

Related Keywords

Canada , Boston , Massachusetts , United States , Ada Lovelace , Zachary Siegel , Belindaz Li , Jacob Andreas , Pratyusha Sharma , Joshua Tenenbaum , Noa Korneev , Julie Shah , Ilia Sucholutsky , Theox Olausson , Jaihai Feng , Maddy Bowers , Madison Robert Hawkins , Jiayuan Mao , Muxin Liu , Thomasl Griffiths , Department Of Aeronautics , University Of California At Berkeley , Natural Sciences , Intel , Harvey Mudd College , Engineering Research Council Of Canada , Us Department Of Defense , Artificial Intelligence Laboratory , Us Office Of Naval Research , Disney , Princeton University , Us National Science Foundation , International Conference On Learning Representations , Watson Ai Lab , Us Defense Advanced Research Projects Agency , University Of Wisconsin , Microsoft , International Conference , Learning Representations , Gabe Grand , Lio Wong , Mini Minecraft , Boston Dynamic Spot , Wisconsin Madison Robert Hawkins , Force Office , Scientific Research , Advanced Research Projects Agency , Naval Research , Science Foundation , Open Philanthropy , Engineering Research Council ,