Marcus Hutter

Understanding Prompt Tuning and In-Context Learning via Meta-Learning

RL, but don't do anything I wouldn't do

Llamas (mostly) think in English: On Causal Interventions in the Latent Language of Transformers

Language Modeling Is Compression

Distributional Bellman Operators over Mean Embeddings

Policy Gradient without Boostrapping via Truncated Value Learning

Neural Networks and Solomonoff Induction

Generative Reinforcement Learning with Transformers