English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Large language models (LLMs) are being applied as actors for sequentialdecision making tasks in domains such as robotics and games, utilizing theirgeneral world knowledge and planning abilities. However, previous work doeslittle to explore what environment state information is provided to LLM actorsvia language. Exhaustively describing high-dimensional states can impairperformance and raise inference costs for LLM actors. Previous LLM actors avoidthe issue by relying on hand-engineered, task-specific protocols to determinewhich features to communicate about a state and which to leave out. In thiswork, we propose Brief Language INputs for DEcision-making Responses (BLINDER),a method for automatically selecting concise state descriptions by learning avalue function for task-conditioned state descriptions. We evaluate BLINDER onthe challenging video game NetHack and a robotic manipulation task. Our methodimproves task success rate, reduces input size and compute costs, andgeneralizes between LLM actors.

Selective Perception: Optimizing State Descriptions with Reinforcement Learning for Language Model Actors