Grounding in AI
Mohit discusses the concept of grounding in AI, emphasizing its importance in understanding coreference through multimodal projects. He shares insights from his early work, where he explored how textual descriptions can enhance the interpretation of 3D images, leading to improved predictions in both text and visual domains. The conversation also touches on the evolution of language models, highlighting the need for models to predict visual mappings alongside textual understanding to address ambiguity in language.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Unifying Vision and Language Models with Mohit Bansal - 636
Related Questions