🧵[1/8]
Excited to share our NeurIPS 2025 Spotlight paper “Does Object Binding Naturally Emerge in Large Pretrained Vision Transformers?” ✨
To add to the broader discussion of binding in neural networks, we ask whether and how Vision Transformers perform object binding (the ability to bind an object’s many features together as a coherent whole).💡
📄 Paper:
[2/8] The binding problem has challenged researchers for decades. If solved, it would enable models to reason efficiently and compositionally at the level of entities or concepts. But modern models still fail here: e.g., diffusion models mixing up attributes and generating a “blue cat” even when asked to produce only red cats and blue dogs. 🔴🐈+🔵🐕➡️🔵🐈
![🧵[1/8]
Excited to share our NeurIPS 2025 Spotlight paper “Does Object Binding Naturally Emerge in Large Pretrained Vision Transformers?” ✨
To add to the broader discussion of binding in neural networks, we ask whether and how Vision Transformers perform object binding (the ability to bind an object’s many features together as a coherent whole).💡
📄 Paper:](/_next/image?url=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FG7DcpP_XAAAJnTL.jpg&w=3840&q=75 "🧵[1/8]
Excited to share our NeurIPS 2025 Spotlight paper “Does Object Binding Naturally Emerge in Large Pretrained Vision Transformers?” ✨
To add to the broader discussion of binding in neural networks, we ask whether and how Vision Transformers perform object binding (the ability to bind an object’s many features together as a coherent whole).💡
📄 Paper:")
