Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to preprocess datasets and train transformer models using PyTorch in this comprehensive tutorial covering Hugging Face Datasets, sentence pair preprocessing, dynamic padding techniques, and training implementation. Master the fundamentals of working with the Datasets library for efficient data handling, understand how to properly preprocess sentence pairs for transformer models, and explore dynamic padding strategies to optimize batch processing. Discover how to leverage the Trainer API for streamlined model training, implement custom training loops in PyTorch, and enhance performance using Hugging Face Accelerate for distributed and mixed-precision training across multiple devices.
Syllabus
Hugging Face Datasets overview (Pytorch)
Preprocessing sentence pairs (PyTorch)
What is dynamic padding?
The Trainer API
Write your training loop in PyTorch
Supercharge your PyTorch training loop with Accelerate
Taught by
Hugging Face