tinyworldformer
An open-source project aimed at running extremely small world models on microcontrollers such as the Raspberry Pi Pico 2 W for real-time generation of game frames, demonstrating the feasibility of extreme edge AI.
Timeline (last 90 days)
The open-source project tinyworldformer runs a miniature world model on a Raspberry Pi Pico 2 W microcontroller and drives a game display.
The tinyworldformer project has received 267 points and 18 comments on Hacker News.
tinyworldformer can run the game in real time on the Pico 2 W, with inference latency kept within an acceptable frame rate range.
tinyworldformer is released as open source on GitHub, where developers can directly view the model structure, training scripts and deployment code.
tinyworldformer performs end-to-end fitting for a single game, compressing the state space into an extremely small discrete set so that the model can degrade into a small state transition table or a shallow network, which is the fundamental reason it can fit into KB-level memory.
All Facts (5)
The open-source project tinyworldformer runs a miniature world model on a Raspberry Pi Pico 2 W microcontroller and drives a game display.
50%UnverifiedThe tinyworldformer project has received 267 points and 18 comments on Hacker News.
50%Unverifiedtinyworldformer can run the game in real time on the Pico 2 W, with inference latency kept within an acceptable frame rate range.
50%Unverifiedtinyworldformer performs end-to-end fitting for a single game, compressing the state space into an extremely small discrete set so that the model can degrade into a small state transition table or a shallow network, which is the fundamental reason it can fit into KB-level memory.
50%Unverifiedtinyworldformer is released as open source on GitHub, where developers can directly view the model structure, training scripts and deployment code.
50%