Explore the intricacies of training large language models in this 1 hour 15 minute podcast episode featuring Databricks' Engineering Manager Bandish Shah and Research Scientist Davis Blalock. Delve into the challenges of scaling language models, from technical hurdles to practical implementation strategies. Gain insights on building effective solutions, optimizing model performance, and ensuring data quality at scale. Learn from the experts' experiences in pushing LLM training limits and helping customers deploy models in production environments. Discover contrarian perspectives on the future of efficient models and uncover valuable lessons for navigating the complex landscape of AI development.
The Art and Science of Training Large Language Models
Overview
Syllabus
[] Bandish and Davis preferred coffee
[] Takeaways
[] Please like, share, leave a review, and subscribe to our MLOps channels!
[] Shout out to everyone who reached out about a paper!
[] AI Quality In-person MLOps Community Conference on June 25th!
[] Shout out to Databricks for sponsoring this episode!
[] Davis' newsletters/ Davis summarizes papers
[] Davis' background in tech
[] Binarization of Neural Networks
[] Bandish's background in tech
[] Day to day life of Davis and Bandish
[] Model Training Challenges
[] Model Consistency Challenges
[] GPU Excess Utilization Efficiency
[] Consistency and Quality Assurance
[] Model Optimization Strategies
[] Simplify, Understand, Improve, Success
[] Complexities of Data
[] Data Quality at Scale
[] Wrap up
Taught by
MLOps.community
Related Courses
-
Training LLMs: Lessons from the Trenches
-
Managing Data for Effective GenAI Application - MLOps Podcast #216
-
Training Large Language Models on Kubernetes
-
Lessons Learned from Scaling Large Language Models in Production
-
MegaScale - Scaling Large Language Model Training to More Than 10,000 GPUs
-
Efficient Large-Scale Language Model Training on GPU Clusters
