MLxDL
A collection of Jupyter notebooks covering agentic AI, multi-agent orchestration, world models and domain-specific LLM fine-tuning, published alongside dozens of models on Hugging Face.
MLxDL is a running collection of notebooks exploring several strands of applied machine learning: autonomous multi-agent demos built with LangGraph, CrewAI and several model providers; implementations of Joint-Embedding Predictive Architecture (V-JEPA and LEJEPA) for latent-space world models; and fine-tuned language models applied to specific domains such as radiology and oncology question answering, flight planning, and market analysis.
The notebooks are the source for dozens of models the author has published on Hugging Face, and the repository recommends viewing them through NBViewer rather than GitHub's own notebook preview, which the README says can time out or crash on the larger, more structurally complex notebooks.
It is written primarily as Jupyter notebooks and licensed under MIT.