LLMs in Production - Fine-Tuning, Scaling, and Evaluation

Dive into the practicalities of deploying Large Language Models (LLMs) in business settings through this 42-minute conference talk by Databricks. Explore when to leverage LLMs and learn how to minimize problem complexity. Gain insights on designing evaluation methodologies and understand when fine-tuning is necessary for optimal performance. Delve into the nuances of training data selection, establishing flexible training ecosystems, hyperparameter optimization, scalable training, and fine-tuning workflows. Follow along with practical demonstrations of the ETL process, data formatting and structuring for fine-tuning, and dataset organization, saving, and management. Witness fine-tuning configurations in action, learn how to monitor and evaluate fine-tuned LLMs, and discover strategies for collecting additional datasets to improve your models over time. Access supplementary resources like the LLM Compact Guide and Big Book of MLOps for further exploration.