Caglar Gulcehre

Principal Researcher@Microsoft·英国·OpenReview

研究方向

large language models · foundation models · multiagent deep reinforcement learning · reinforcement learning · imitation learning · demonstrations · attention models · deep learning · nlp · natural language understanding · optimization · cognitive science · cognitive neuroscience

Caglar Gulcehre

From Markov to Laplace: How Mamba In-Context Learns Markov Chains

The Diffusion Duality, Chapter II: $\Psi$-Samplers and Efficient Curriculum

Partition Generative Modeling: Masked Modeling Without Masks

Control Tax: The Price of Keeping AI in Check

Adaptive Attacks on Trusted Monitors Subvert AI Control Protocols

Loopholing Discrete Diffusion: Deterministic Bypass of the Sampling Wall

Quantile Reward Policy Optimization: Alignment with Pointwise Regression and Exact Partition Functions

Algorithm Discovery With LLMs: Evolutionary Search Meets Reinforcement Learning

Beyond Autoregression: Fast LLMs via Self-Distillation Through Time

One-Step is Enough: Sparse Autoencoders for Text-to-Image Diffusion Models

RAT: Bridging RNN Efficiency and Attention Accuracy via Chunk-based Sequence Modeling

Fleet of Agents: Coordinated Problem Solving with Large Language Models

SIKeD: Self-guided Iterative Knowledge Distillation for Mathematical Reasoning

Unpacking SDXL Turbo: Interpreting Text-to-Image Models with Sparse Autoencoders