PinnedTechnocratinCoderHack.comGlore: Memory-Efficient LLM Training by Gradient Low-Rank ProjectionGlore (Gradient Low-Rank Projection) is a memory-efficient technique for training large language models (LLMs) from scratch using a single…·3 min read·Mar 12, 2024----
PinnedTechnocratinCoderHack.comGitHub Actions for CI/CDContinuous integration (CI) and continuous delivery/deployment (CD) are practices that are intended to minimise mistakes when making…·6 min read·Sep 13, 2023----
TechnocratinDev Geniuspolars library: fast dataframe for curious data scientists·3 min read·May 30, 2024----
TechnocratinCoderHack.compolars library: fast dataframe for curious data scientists·4 min read·Apr 24, 2024----
TechnocratinCoderHack.comGrok model — first readRecently released grok1 at https://github.com/xai-org/grok-1/blob/main/model.py·3 min read·Mar 18, 2024----