Document Receptive Fields
Yannic and Connor discuss the challenges of inputting multiple documents into transformer models, exploring the impact of document length on model performance and the potential for cross-document attention within transformer layers.In this clip
From this podcast
Machine Learning Street Talk (MLST)
OpenAI GPT-3: Language Models are Few-Shot Learners
Related Questions