5 papers · 1 filter
Learning Situation-Conditioned Thinking Policies for Long-Term LLM Agents
Hong Su
Long-running autonomous agents must reuse accumulated reasoning experience without allowing explicit historical memory and LLM context to grow indefinitely. However, existing memor…
Theoretical guarantees for stochastic gradient Langevin dynamics
Daniel Paulin, Peter A. Whalley
We prove asymptotic bias bounds for stochastic gradient Langevin dynamics in Wasserstein distance of order two. We assume that the negative log-density is strongly convex with a Li…
Training-Free Generative Sampling via Moment-Matched Score Smoothing
Zhenyu Yao, Daniel Paulin
Diffusion models generate samples by denoising along the score of a perturbed target distribution. In practice, one trains a neural diffusion model, which is computationally expens…
Theoretical guarantees for stochastic gradient sampling methods via Gaussian convolution inequalities
Daniel Paulin, Peter A. Whalley
We derive first-order (in the stepsize) bounds on the bias in Wasserstein distances of the invariant measure of stochastic gradient kinetic Langevin dynamics with minimal assumptio…
Infinite-dimensional generative diffusions via Doob's h-transform
Thorben Pieper-Sethmacher, Daniel Paulin
This paper introduces a rigorous framework for defining generative diffusion models in infinite dimensions via Doob's h-transform. Rather than relying on time reversal of a noising…