Reformer - The Efficient Transformer
Offered By: Yannic Kilcher via YouTube
Course Description
Overview
Explore the groundbreaking Reformer model in this informative video, which addresses the resource-intensive nature of the famous Transformer architecture. Learn how the Reformer combines Locality Sensitive Hashing and concepts from Reversible Networks to significantly reduce memory usage and enable processing of much longer input sequences. Discover how this innovative approach allows for handling up to 16K tokens with just 16GB of memory, making it a game-changer for natural language processing tasks. Delve into the technical details of the model's O(LlogL) complexity, reversible residual layers, and their impact on efficiency. Gain insights into the Reformer's performance, which rivals traditional Transformer models while offering substantial improvements in memory efficiency and processing speed for long sequences.
Syllabus
Reformer: The Efficient Transformer
Taught by
Yannic Kilcher
Related Courses
Mining Massive DatasetsStanford University via edX Building Features from Text Data
Pluralsight Locality Sensitive Hashing for Search with Shingling + MinHashing - Python
James Briggs via YouTube Private Nearest Neighbor Search with Sublinear Communication and Malicious Security
IEEE via YouTube Time Signature Based Matching for Data Fusion and Coordination Detection in Cyber Relevant Logs
0xdade via YouTube