AI After Hours | Wednesday 19 November
Why VLMs cannot parse your documents reliably
Machine Learning Researchers, Evan Vogelbaum and Yifei Hu, from Reducto break down the strategy for achieving best-in-class document processing accuracy. Discover the blended approach Reducto uses, combining the reliability of traditional OCR with the contextual understanding of Vision Language Models (VLMs). Learn why naively using VLMs for documents can lead to failures like hallucination and high latency, and how Reducto engineered a pipeline to mitigate these issues.
Speakers
Evan Vogelbaum
ML Researcher @ Reducto
Yifei Hu
ML Researcher @ Reducto
About AI After Hours
Deep Learning Leaders provides a platform for the world's foremost thinkers in AI to talk about the current challenges, future opportunities and their journeys in the industry. Guests have included Luc Vincent, the creator of Google Street view (a product now with over 1 billion MAU), and Victor Riparbelli, CEO Synthesia (genAI Unicorn).
AI After Hours
Upcoming Events
AI After Hours London: Holiday Edition
AI builders - join Encord for our Holiday Edition of our AI After Hours speaker series. There'll be festive drinks, appetizers, and insightful discussions. Co-hosted with MMC Ventures in London.
Upcoming Events
AI After Hours London: Holiday Edition
AI builders - join Encord for our Holiday Edition of our AI After Hours speaker series. There'll be festive drinks, appetizers, and insightful discussions. Co-hosted with MMC Ventures in London.