Apache Cassandra Lunch 139: LLM Fine-Tuning with QLoRA - Evaluation

Welcome to another insightful session of our Cassandra Lunch series, where we delve deeper into the intricacies of Large Language Model (LLM) fine-tuning. This episode is dedicated to unraveling the critical process of evaluating LLM performance, a cornerstone in the journey of fine-tuning your models to perfection.

After fine-tuning the model (see our talk from 2 weeks ago), LLMs demand a more nuanced approach to assessment. We will shed light on a few methods for evaluating the performance of fine-tuned LLMs compared to the responses before fine-tuning. These tailored metrics provide deeper insights, helping to ensure that your fine-tuning efforts align precisely with your model's intended goals and applications.

Whether you are a seasoned practitioner or new to the world of LLM fine-tuning, this session is designed to equip you with the knowledge and tools needed to effectively evaluate and refine your language models.

Join us as we navigate the complexities of LLM evaluation, paving the way for enhanced performance and groundbreaking advancements in generative AI.

Bring your lunch and join in. Don't have to leave your desk.

5m Intro.
25-50m Volunteer presents/ talks about something they are working on/cool stuff
5-15m Q/A Commentary