This article explores how Retrieval Augmented Generation (RAG) can be used with large language models to suggest recipes based on user-provided ingredients. It walks through building a recipe recommendation system using embeddings, vector search, and prompt engineering, complete with code examples and a simple web interface.
The post describes creating a FastAPI endpoint to return recipes based on input ingredients, using a prebuilt model and saved artifacts. It covers setup, artifact loading, sanity checks, and cluster-to-index mappings for quick lookups.
This post introduces unsupervised learning, focusing on clustering and its use for grouping recipes by ingredient similarity. It describes vectorizing ingredients with TF-IDF and applying k-means to form clusters.
We compare a compact fine-tuned LLM against a lightweight supervised model for parsing recipes from noisy web pages. The article outlines experiments with commercial LLMs, local open-source models, and cost-performance considerations.
This article discusses speeding up ML training loops by caching serialized features to reduce repeated data processing time. It covers loading cached features, memory profiling, and evaluating a Logistic Regression model with performance…