Data and Model Sharing
Jeremy highlights the challenges of accessing locked datasets in NLP, emphasizing the need for more accessible data sharing. He advocates for the release of pre-trained models in medical imaging, suggesting that sharing model weights can lead to equally effective results as sharing raw data. This collaborative approach could enhance the development of specialized models across institutions.In this clip
From this podcast

The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Trends in Deep Learning with Jeremy Howard - TWiML Talk #214
Related Questions