VVlad Butacu·Apr 15Your Local LLM Is Slow Because of Five Config FlagsKV cache, context size, and batch size control more of the performance budget than most people realize8 min read·omniforge.online
KKashif·Feb 8I Pretrained a 360M LLaMA-Style Language Model from Scratch on 6B FineWeb Tokens (Single H100)Explore pretraining a 360M LLaMA-style language model on 6B tokens using a single H100 GPU, with reproducible results5 min read·blog.ifkash.dev
VVenkat Raman·Apr 18, 2024The power of Mechanical Sympathy in Software EngineeringExploring Locality of Reference, LMAX Disruptor & Flash Attention22 min read·venkat.eu