← ALL TAGS

Tag: apple-silicon

4 articles tagged “apple-silicon” on Oak Oliver Engineering, newest first. The latest is “Sounds and Videos as Prompts: A 107M Adapter That Lets EmbeddingGemma 2 Drive Qwen-Image and Iris, Retrained on Permissive Data”.

Engineering

Sounds and Videos as Prompts: A 107M Adapter That Lets EmbeddingGemma 2 Drive Qwen-Image and Iris, Retrained on Permissive Data

We replaced the 8B text encoder of Qwen-Image 2.1 with the 0.74B EmbeddingGemma 2 plus a small adapter, so a sound or a short video can be referenced in a prompt like an image. Then we threw away every non-commercial training item and retrained on CC0 / CC BY data only. The permissive adapter is better on text, the best we have on video, and the licensing surprise was not in the data.

October 21, 2026 11 min read
Read Story →
Engineering

Oak Oil: Cargo Plans, We Execute — Lean, Fast Rust Builds That Stay Byte-Identical to Stock Cargo

One Mac held 469 GiB of Cargo target directories, 177 GiB of them byte-identical copies. Oak Oil borrows Bun's build architecture: Cargo plans, our executor runs the plan against a content-addressed store. Output is byte-identical to stock Cargo, a wiped target comes back in 1.7 s instead of 11 s, and cleanup is verified by Cargo itself. Open source, built and published in public.

October 6, 2026 10 min read
Read Story →
Engineering

Porting TRELLIS.2 From Python to Rust and Metal: 2D Sprites to 3D Models in One Command

I moved Microsoft's 4-billion-parameter image-to-3D pipeline off PyTorch and onto our Rust and Metal runtime. One command now turns a 2D sprite into a textured GLB in one to three minutes with about 6 GB of memory instead of 33. This is how the port was organised, the three bugs worth retelling, the sparse-convolution kernel that decided it, and what parity means when the sampler is chaotic.

October 5, 2026 12 min read
Read Story →
Engineering

Porting ERNIE-Image-Turbo to Zig and Metal: An 8B Diffusion Transformer From Scratch on Apple Silicon

I rewrote the hot path of Baidu's 8B-parameter ERNIE-Image-Turbo (DiT + scheduler) from scratch in Zig with hand-written Metal kernels. 13.06s/step at 1024×1024 on M2 Max. The bug archaeology behind getting 36 layers numerically correct, plus the honest accounting of what still has Python in it.

April 23, 2026 18 min read
Read Story →