latent-variable

LeAct introduces a method to recover chain-of-thought reasoning from silent expert actions by treating the CoT as a latent variable and retaining only those that improve the student's probability of reproducing the action, outperforming baselines in games and robotics.

This paper proposes a method to decompose the response distribution of language models into structured, strategy-conditioned representations using a latent variable, addressing posterior collapse with a model-directed reconstruction objective.

This paper introduces Generative Recursive reAsoning Models (GRAM), a probabilistic framework that extends recursive reasoning models by enabling stochastic latent trajectories, multiple hypotheses, and inference-time scaling through depth and parallel sampling.