Summary of Playing Nethack with Llms: Potential & Limitations As Zero-shot Agents, by Dominik Jeurissen and Diego Perez-liebana and Jeremy Gow and Duygu Cakmak and James Kwan
Playing NetHack with LLMs: Potential & Limitations as Zero-Shot Agents
by Dominik Jeurissen, Diego Perez-Liebana, Jeremy Gow, Duygu Cakmak, James Kwan
First submitted to arxiv on: 1 Mar 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents NetPlay, a Large Language Model (LLM) powered zero-shot game-playing agent that successfully navigates the challenging roguelike game NetHack. While LLMs have shown promise as high-level planners for agents in simpler environments like Minecraft, NetHack’s diverse set of items and monsters, complex interactions, and many ways to die pose a significant challenge. The authors leverage the power of LLMs to develop an agent that can learn to plan and make decisions without explicit training data. By evaluating the performance of NetPlay on NetHack, this paper contributes to the advancement of LLM-based agents in dynamic environments. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary NetPlay is a special kind of computer program that helps other programs make good decisions without needing to be trained beforehand. This technology has been tested in simple games like Minecraft and worked well. But what about more complex games? The creators of NetPlay decided to test it on a game called NetHack, which is super hard because there are many things you can do or collect that affect the game in different ways. They wanted to see if their program could still make good decisions even with all these possibilities. |
Keywords
» Artificial intelligence » Large language model » Zero shot
