rafaelvp-db/databricks-llm-prompt-engineering

Combines instruction-tuned models (MPT-7B, LLaMA2) with MLflow's LLMOps evaluation tools (`mlflow.evaluate()`) and Databricks Model Serving for GPU-accelerated inference, enabling end-to-end experimentation from prompt optimization through production deployment. Includes performance optimization patterns like vLLM for 7-10X latency improvements and demonstrates active prompting with chain-of-thought techniques. Ships with a Gradio frontend for interactive testing against deployed Model Serving endpoints.

No commits in the last 6 months.