# Show HN: I built a playable 211M-param world model of the Backrooms

> Source: <https://github.com/Yelar/noclip>
> Published: 2026-11-21 22:00:20+00:00

**Key takeaway:** A 211M-parameter transformer can serve as a game engine by predicting each frame from the last 2.4 seconds of frames plus inputs, trained on 440 hours of scripted play and runnable on an NVIDIA GPU with about 4 GB of memory.

A 211M-param transformer acts as a game engine. It draws every next frame from the last 2.4 seconds and my inputs.

Built a simulator of 18 Backrooms levels in PyTorch to generate the data.

Recorded 440 hours of scripted play in it.

Trained the model from scratch on that, using flow matching in a pretrained autoencoder's latent space.

The model was trained on its own imperfect frames as context so it doesn't drift right away. It recalls a few older frames by position, so a quick look back usually works.

Inspired by Oasis Minecraft.

To try it you need an NVIDIA GPU with ~4 GB.

Weights: [https://huggingface.co/yelarys/noclip](https://huggingface.co/yelarys/noclip)

Video with sound: [https://x.com/yelaryss](https://x.com/yelaryss)

Comments URL: [https://news.ycombinator.com/item?id=50594070](https://news.ycombinator.com/item?id=50594070)

Points: 1

# Comments: 0
