Project Malmo

Join me in attending the next episode of Research Forum, a series from Microsoft Research that explores bold ideas, new advances, and important discussions with the global research community in the era of general AI. Register: register.researchforum.microsoft.com/?OCID=msr_rese…

MineLand: Simulating Large-Scale Multi-Agent Interactions with Limited Multimodal Senses and Physical Needs github.com/cocacola-lab/M…

Large Language Models as Minecraft Agents arxiv.org/abs/2402.08392

Describe, Explain, Plan and Select: Interactive Planning with Large Language Models Enables Open-World Multi-Task Agents openreview.net/pdf?id=KtvPdGb…

JARVIS-1: Open-World Multi-task Agents with Memory-Augmented Multimodal Language Models paper page: huggingface.co/papers/2311.05… Achieving human-like planning and control with multimodal observations in an open world is a key milestone for more functional generalist agents. Existing approaches can handle certain long-horizon tasks in an open world. However, they still struggle when the number of open-world tasks could potentially be infinite and lack the capability to progressively enhance task completion as game time progresses. We introduce JARVIS-1, an open-world agent that can perceive multimodal input (visual observations and human instructions), generate sophisticated plans, and perform embodied control, all within the popular yet challenging open-world Minecraft universe. Specifically, we develop JARVIS-1 on top of pre-trained multimodal language models, which map visual observations and textual instructions to plans. The plans will be ultimately dispatched to the goal-conditioned controllers. We outfit JARVIS-1 with a multimodal memory, which facilitates planning using both pre-trained knowledge and its actual game survival experiences. In our experiments, JARVIS-1 exhibits nearly perfect performances across over 200 varying tasks from the Minecraft Universe Benchmark, ranging from entry to intermediate levels. JARVIS-1 has achieved a completion rate of 12.5% in the long-horizon diamond pickaxe task. This represents a significant increase up to 5 times compared to previous records. Furthermore, we show that JARVIS-1 is able to self-improve following a life-long learning paradigm thanks to multimodal memory, sparking a more general intelligence and improved autonomy.

The ToMCAT Dataset, a rich, multimodal dataset consisting of data from 40 teams of three humans conducting simulated urban search-and-rescue (SAR) missions in a Minecraft based testbed, collected for the ToMCAT project. openreview.net/pdf?id=ZJWQfgX…

The #greenlands platform is now live! 🤖🧑‍🤝‍🧑 With the ability to test #AI agents with real humans and receive invaluable feedback, this could be a game-changer for evaluation methods! Don't miss out, check it out now! #artificialintelligence #gaming #evaluation #Minecraft

Students can learn Azure via server creation, dive into #AI using Minecraft & GPT3 OpenAI or learn with Minecraft using Microsoft Research Project Malmo. @wongcyrus shares how to set up a free private Minecraft server in Azure via #EDU yearly credit. bit.ly/3Fh3CHC

Nvidia AI plays Minecraft, wins machine&learning conference award: NeurIPS 2022 honors MineDojo for playing Minecraft when instructed by written prompts. dlvr.it/SdYpHf #technews #ev #tesla

📢📢We are happy to announce the winners IGLU NLP track at @NeurIPSConf: 🥇First place – Craftsmanfly team: zubingou, wxl1999, guo_qingyan, jason_zhang 🥈Second place – try1try team 🥉Third place – felipe_b team: Felipe Bivort Haiek aicrowd.com/challenges/neu…

Happy to share our job post for *in person* summer internships! careers.microsoft.com/us/en/job/1488…

📢It's never too late! There is still one month left to the IGLU 2022 @NeurIPSConf NLP task deadline. Get your hands dirty by playing with our baseline. "What to ask as Clarifying questions?" 🎁 $9,000 cash prize + Co-authorship #NeurIPS2022 #NLP #AI bit.ly/3y5uYgS

Heading to London (for the first time in over 2 years🤯) to talk @ic_arl @ICComputing about Human-AI collaboration for improved player experience today at 5pm (BST). Free registration for in-person or remote is still open at: eventbrite.com/e/imperial-col…

You can register now for our next seminar with @smdvln from @MSFTResearch who will be talking about human-AI team-work collaboration!🤾🤖 Sign up here eventbrite.com/e/imperial-col… 📍@imperialcollege Huxley 308 @ICComputing #AI

There’s still time to apply! ✍️

In this recently published article we explore why an AGI-ish Minecraft agent is valuable, what problems it uniquely solves, and why it is valuable for solving complex real-world problems. 👇 🔗 blog.singularitynet.io/toward-general… #SNETCommunity #Minecraft #AI 🧵1/2 ⬇️

Sam Devlin @smdvln taking about designing games for AI to play at #gameaischool2022

Designing Games for AI to Play at #GameAISchool2022 with @smdvln from @MSFTResearch

Congratulations to Mike's Angels, the winners of the 2022 AI Settlement Generation Challenge in @Minecraft. This is a thread about their entry and the GDCM competition. Feel free to share. #pcg #Minecraft #GameAI @UniLeidenNews @mike_preuss

Missing the MineRL Diamond 💎 competition? Don’t worry: it is now the intro track for the BASALT competition… with a twist. You can develop reinforcement and imitation learning agents to get a diamond *shovel*, then move to the BASALT tasks! Start here: aicrowd.com/challenges/neu…