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

The captivating realm of Minecraft has attracted substantial researchinterest in recent years, serving as a rich platform for developing intelligentagents capable of functioning in open-world environments. However, the currentresearch landscape predominantly focuses on specific objectives, such as thepopular "ObtainDiamond" task, and has not yet shown effective generalization toa broader spectrum of tasks. Furthermore, the current leading success rate forthe "ObtainDiamond" task stands at around 20%, highlighting the limitations ofReinforcement Learning (RL) based controllers used in existing methods. Totackle these challenges, we introduce Ghost in the Minecraft (GITM), a novelframework integrates Large Language Models (LLMs) with text-based knowledge andmemory, aiming to create Generally Capable Agents (GCAs) in Minecraft. Theseagents, equipped with the logic and common sense capabilities of LLMs, canskillfully navigate complex, sparse-reward environments with text-basedinteractions. We develop a set of structured actions and leverage LLMs togenerate action plans for the agents to execute. The resulting LLM-based agentmarkedly surpasses previous methods, achieving a remarkable improvement of+47.5% in success rate on the "ObtainDiamond" task, demonstrating superiorrobustness compared to traditional RL-based controllers. Notably, our agent isthe first to procure all items in the Minecraft Overworld technology tree,demonstrating its extensive capabilities. GITM does not need any GPU fortraining, but a single CPU node with 32 CPU cores is enough. This researchshows the potential of LLMs in developing capable agents for handlinglong-horizon, complex tasks and adapting to uncertainties in open-worldenvironments. See the project website at https://github.com/OpenGVLab/GITM.

Ghost in the Minecraft: Generally Capable Agents for Open-World Environments via Large Language Models with Text-based Knowledge and Memory