Unity Muse & Sentis, Model-Based RL, and Diplomacy

Welcome to our weekly newsletter and platform update! Every week we’ll bring you the latest AI + gaming news, resources, and updates from the Regression Games team.

Join our community on Discord and start playing!

Weekly Spotlight - Unity Muse and Sentis

Unity’s highly-anticipated suite of AI tools has finally been introduced this morning via a blog post. Unity Muse, a generative AI platform to create assets and interact with the editor, and Unity Sentis, a product to run ML models directly within the Unity runtime, are slated to be released in the coming months.

Unity is looking towards a future where the creative work of developers can happen via human-language input. In their blog post, they talk about how they want to reduce the reliance on complicated UIs, and instead use tools like chat for a developer to simply tell Unity what to build out, whether it be images, environments, or entirely new components of your experience.

With Unity Sentis, the effort needed to embed models directly within a Unity game, useful for features such as generated NPC dialog, will also become easier.

Regression Games Platform Update

The Regression Games team has been gearing up for some big updates over the coming weeks. We plan to announce these updates via a public roadmap, as well as on our social media accounts and Discord.

In addition to a roadmap for players and our competitive AI platform, we also have exciting updates regarding initial availability of our platform for studios. We are excited to begin testing our bot-building tools for Unity with studios, and we appreciate those who have given us feedback so far! If you are interested in learning more about our product offer, please email [email protected] to get in touch.

Blog Post / Presentation - Model Based Reinforcement Learning

This week’s highlighted blog post is from Isaac Kargar, which covers a large swath of details and topics in the space of model based reinforcement learning, where RL agents use a model of the world to help with completing tasks. The talk is quite deep, but if you are looking to learn a lot about the topic, this talk is for you!

Technical Paper of the Week

Our technical paper this week is an interesting alternative paper to previous work we’ve seen on having agents play Diplomacy, a strategy game. In “Mastering the Game of No-Press Diplomacy via Human-Regularized Reinforcement Learning and Planning” from Meta AI and MIT, the authors cover an approach where agents play Diplomacy without natural language communication, and must communicate via their actions. This was an interesting spin on the game that proved difficult for earlier approaches.

Members of our community discuss these papers on our Discord #reading-club channel. You can also find all past resources on our GitHub.

Looking for something new? Join the Regression Games team!

Have feedback about our newsletter? Let us know at [email protected]

You are receiving this email because you registered to our mailing list or are a player on our platform.