$ show UpstageAI/cookbook
This repository is an educational cookbook from Upstage, centered on their Solar LLM and Document AI APIs. It contains a large collection of Jupyter notebooks (Solar-Fullstack-LLM-101) that teach prompt engineering, RAG, cache-augmented generation, tool calling, and fact-checking with knowledge graphs, plus a Flask sample backend (Bremen) and a Flutter mobile app. The dominant character is a notebook-driven ML/research collection, but it is hybrid because it includes runnable application code (Flask API, Flutter app) and static data assets (OCR JSON, CSV knowledge bases).
The LLM identified 5 files worth highlighting for reuse:
Reusable prompt engineering patterns including role definition, few-shot learning, XML tags, JSON output constraints, and self-verification. Demonstrates how to structure LLM prompts for reliable structured output, applicable to any LLM-based application.
End-to-end RAG with tool calling, showing how to combine retrieval with function calling for grounded generation. A reusable pattern for building knowledge-grounded LLM agents that need up-to-date or domain-specific information.
Cache-augmented generation pattern that reuses context across calls to reduce latency and cost. Demonstrates a non-obvious optimization for repeated LLM queries and is adaptable to other LLM backends.
A simple CSV knowledge base schema (chunk_id, chunk_category, chunk_text) used for the health agent example. This is a clean, portable data format for any domain-specific RAG pipeline and can be reused with minimal adaptation.
Flask backend exposing endpoints for PDF upload, universal extraction, recommendation, and perplexity scoring. Shows how to wrap Upstage Document AI and Solar LLM APIs in a REST service, serving as a reference for building document-processing microservices.