Andrej Risteski

Associate Professor@Carnegie Mellon University·美国·OpenReview

研究方向

deep learning theory · deep generative models · unsupervised learning · out-of-distribution generalization · OOD generalization · probabilistic graphical models · variational inference · markov chain monte carlo · word embeddings

6.5

Taming Imperfect Process Verifiers: A Sampling Perspective on Backtracking

ICLR 2026Poster

3.3

CodePDE: An Inference Framework for LLM-driven PDE Solver Generation

ICLR 2026Withdrawn

3.0

Towards Understanding the Transferability of Adversarial Suffixes in Large Language Models

合作者 (20)

Andrej Risteski

Taming Imperfect Process Verifiers: A Sampling Perspective on Backtracking

CodePDE: An Inference Framework for LLM-driven PDE Solver Generation

Towards Understanding the Transferability of Adversarial Suffixes in Large Language Models

Progressive distillation induces an implicit curriculum

On the Benefits of Memory for Modeling Time-Dependent PDEs

Towards characterizing the value of edge embeddings in Graph Neural Networks

On the Query Complexity of Verifier-Assisted Language Generation

Towards characterizing the value of edge embeddings in graph neural networks