Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn the fundamentals of subword tokenization in this comprehensive lecture that explores how text is broken down into smaller meaningful units for natural language processing tasks. Discover the motivation behind subword tokenization, including handling out-of-vocabulary words and improving model performance across different languages. Examine popular subword tokenization algorithms such as Byte Pair Encoding (BPE), WordPiece, and SentencePiece, understanding their mechanisms and trade-offs. Explore practical applications in modern NLP models like BERT, GPT, and other transformer architectures. Gain hands-on insights into implementation considerations, vocabulary size selection, and the impact of different tokenization strategies on downstream tasks. Access accompanying slides to reinforce key concepts and follow along with detailed examples throughout this hour-long deep dive into one of the foundational preprocessing steps in contemporary natural language processing.
