MIRA - Blog post

Multiplayer world model - a dream of Rocket League.

MIRA: Multiplayer Interactive World Models with Representation Autoencoders# Hover/tap on one quarter to zoom. We train a model to simulate Rocket League, Epic Games' car-football game. It allows four players to play a 2v2 match together, responding in real time to their actions and generating a consistent view of the game at 20 fps, with a resolution of 576p split across the four players. It's a 5B-parameter diffusion transformer paired with a 600M-param video representation codec. We're open-sourcing the dataset as well as training and inference code. The project is a stepping stone to…

saved by

• Eric R

related reading

• Scaling Video Pretraining with Imagination Modelsinductionlabs.com
• How to train a frontier-level world modelnext-state.github.io
• How not to do research - Rajan Agarwalrajan.sh
• The First Fully General Computer Action Model | blogsi.inc
• World Models | Rohit Bandarurohitbandaru.github.io
• Deep Reinforcement Learning: Pong from Pixelskarpathy.github.io
• LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixelsle-wm.github.io
• World Models: Computing the Uncomputablenotboring.co
• Are Video Generation Models World Simulators? · Artificial Cognitionartificialcognition.net
• RTFM: A Real-Time Frame Model | World Labsworldlabs.ai
• A General Goal-Conditioned Minecraft Model - Pantographpantograph.com
• [2603.08546] Interactive World Simulator for Robot Policy Training and Evaluationarxiv.org