PyGlove is Google’s library for symbolic programming in Python, enabling manipulation of programs as mutable objects for AutoML, hyperparameter tuning, and neural architecture search.
Autofaiss automatically selects the best Faiss index type and parameters for your embeddings, making it easy to build fast and memory-efficient approximate nearest neighbor search indexes.
Gin-config is Google’s lightweight configuration framework that uses decorators and .gin files to make Python functions and classes configurable without boilerplate code.
Learn how BERTScore evaluates generated text with contextual embeddings, when to use it instead of ROUGE, BLEU, or LLM-as-judge, and how to wire it into watsonx.ai evaluation workflows.
How Ollama makes local LLMs practical: setup, the Python client and OpenAI-compatible API, when local inference beats hosted models, and how it pairs with watsonx.ai in hybrid architectures.
A practical guide to Folium for interactive Leaflet.js maps in Python: markers, choropleths, heatmaps, when to choose it over Plotly or pydeck, and how to publish maps from data pipelines.
A practical guide to LangGraph, the library for building stateful, multi-actor LLM applications using graph-based workflows with cycles, branching, and persistence.