Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore ByteDance's innovative approach to building large-scale data processing pipelines for multimodal models using Ray in this 35-minute conference talk. Discover how Xiaohong Dong, Wanxing Wang, and Liguang Xie from ByteDance tackled the challenges of processing vast amounts of high-quality video data for advanced video generation models. Learn about their utilization of Ray's ecosystem, including Ray Core, Ray Data, and Ray Serve, to create a robust and scalable data pipeline. Gain valuable insights into managing Ray infrastructure, best practices for large-scale multimodal AI projects, and solutions for dynamic scaling and orchestration of heterogeneous resources. Uncover a blueprint for leveraging Ray in ambitious AI endeavors and understand how ByteDance overcame the complexities of handling massive video datasets.
