Stormin' The Castle

Archive

  1. Software Factories
  2. AI will be your only UI
  3. VoiceDroid - A Fast, Voice-Driven Computer-Use Agent for Your Phone
  4. Learning Position — Trainable Embeddings from BERT to GPT
  5. Understanding Positional Encodings in Transformers
  6. Claude Code Power Tips
  7. Sparse Attention
  8. Limitless Video
  9. Still Talking?
  10. Tesla FSD
  11. Autostart Node.js Apps at Boot
  12. Finetuning Redpajama (OpenLlama)
  13. PilotWatch
  14. miniSD - Fast Quarter-Sized Image Generation using Stable Diffusion
  15. Alpaca Finetuning of Llama on a 24G Consumer GPU
  16. Object Detection - Part 5 Final
  17. Tiny ChatGPT
  18. Object Detection - Part 4 Multiple Objects
  19. Posing Heads with Stable Diffusion
  20. Object Detection - Part 3 Compound Loss
  21. Object Detection - Part 2 Bounding Boxes
  22. Object Detection from Scratch - Part 1
  23. DiffEdit - Semantic Image Editing with Stable Diffusion
  24. UFO Lands on Highway! Or Depth Estimation using ML
  25. Catch! Using Recurrent Visual Attention.
  26. Pong From Pixels.
  27. Video Search Engine Using OpenAI's CLIP
  28. Unleash the GPU in your Colab Notebook
  29. Relight.

Projects

Sandtoy
Sandtoy
Fun little pixel toy written in WebGL using a fragment shader to do GPU accelerated particle physics.
Stable Diffusion Poser
Stable Diffusion Poser
Control Stable Diffusion Image Generation with an Image Regression Model for Head Pose.
Depth Viewer
Depth Viewer
A Hugging Face Space fusing a machine learning model for estimating depth from pixels and a WebGL 3d depth viewer.
Picture Pile
Picture Pile
Little WebGL image viewer incorporating animation and non-linear interpolation.

Share on Twitter 

John Robinson © 2022-2026