← All repos

yousef-rafat/MaximusLLM

View on GitHub ↗ ★ 28 Python archived

High-throughput long-context LLMs. Scaling context via RandNLA and massive vocab capacity through MAXIS Loss and Fisher-SVD.

4 mentions on Hacker News by 1 person: 4 stories, 0 comments · first 2026-03-13 · latest 2026-03-16