Yann LeCun

13

Papers

991

Total Citations

Papers (13)

Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs

Navigation World Models

Layer by Layer: Uncovering Hidden Representations in Language Models

Transformers without Normalization

Scaling Language-Free Visual Representation Learning

Hierarchical World Models as Visual Whole-Body Humanoid Controllers

RoboPEPP: Vision-Based Robot Pose and Joint Angle Estimation through Embedding Predictive Pre-Training

Seq-VCR: Preventing Collapse in Intermediate Transformer Representations for Enhanced Reasoning

Rate-In: Information-Driven Adaptive Dropout Rates for Improved Inference-Time Uncertainty Estimation

MetaMorph: Multimodal Understanding and Generation via Instruction Tuning

How Learning by Reconstruction Produces Uninformative Features For Perception

The Entropy Enigma: Success and Failure of Entropy Minimization

Stochastic positional embeddings improve masked image modeling