Stories by yousef_g

Show HN: MaximusLLM – Train 262k-vocab LLMs on a single 16GB GPU

Ghost Logits: Simulating missing partition mass in sampled softmax [pdf]

Show HN: MaximusLLM, Breaking transformer's O(N^2) and O(V) scaling bottlenecks

MaximusLLM: High-Speed Architecture via Ghost Logits and Random Latent Attention

Self-Adapting Language Models

I have reimplemented Stable Diffusion 3.5 from scratch in pure PyTorch

Deepseek-R1 32B Training with Supervised Fine-Tuning

Ask HN: Any DIY Hardware tutorials out there?

Magna: Embedding similarity search tool for searching within large documents

RustyChat: Asynchronous local chat server written in Rust

Networking Tutorial Playlist by Ben Eater

Open Source Twitter Bot