Explore ByteDance's innovative approach to building a video data processing pipeline for reinforcement learning from human feedback in this Ray Summit 2024 presentation. Gain insights into leveraging Ray's ecosystem to develop a scalable system capable of producing realistic video scenes from text instructions. Learn how Ray Core, Ray Data, and Ray Serve were utilized to handle massive volumes of video data efficiently. Discover the importance of dynamic computation scaling and heterogeneous resource orchestration in constructing complex data processing systems. Examine ByteDance's methods for managing Ray infrastructure and acquire best practices for implementing large-scale data processing solutions. This talk is particularly valuable for teams working on AI models requiring extensive multimedia dataset processing and analysis.
ByteDance's Platform for Reinforcement Learning from Human Feedback
Overview
Syllabus
ByteDance's Platform for Reinforcement Learning from Human Feedback | Ray Summit 2024
Taught by
Anyscale
Related Courses
-
Reinforcement Learning from Human Feedback (RLHF)
-
Reinforcement Learning from Human Feedback (RLHF)
-
RLHF: How to Learn from Human Feedback with Reinforcement Learning
-
Distributed Mesh Processing and Geometric Deep Learning for Wind Prediction
-
Reinforcement Learning from Human Feedback (RLHF) - Advances in Generative AI
-
Ray/RLLib-Empowered Reinforcement Learning Based Recommender Systems in NetEase Game
