Accelerating Data Intensive ML Applications with the Graphcore IPU

Explore how Graphcore accelerates AI development through their Intelligence Processing Unit (IPU) hardware and user-friendly examples in this 55-minute conference talk from the Data Science Festival. Discover the implementation and optimization of BERT-Large for IPU systems, showcasing state-of-the-art performance results using industry-standard machine learning training schemes. Learn to access IPUs via Spell's Cloud MLOps platform, follow a BERT Fine-tuning notebook tutorial using the SQuADv1 dataset, and run inference question-answering tasks with the HuggingFace inference API. Gain insights into training transformer models faster using the Hugging Face Optimum toolkit, and understand how Graphcore's technology can benefit industries like legal, banking, finance, and healthcare undergoing significant AI transformation.

Accelerating data intensive ML applications with the Graphcore IPU - Data Science Festival