9 papers
Temper-Then-Tilt: Principled Unlearning for Generative Models through Tempering and Classifier Guidance
Jacob L. Block, Mehryar Mohri, Aryan Mokhtari +1
We study machine unlearning in large generative models by framing the task as density ratio estimation to a target distribution rather than supervised fine-tuning. While classifier…
Entropy Aware Reward Guidance for Diffusion Language Model Alignment
Atula Tejaswi, Litu Rout, Constantine Caramanis +2
Reward guidance, also known as posterior sampling, is a popular method for test-time adaptation and post-training in continuous diffusion models. In this paper, we study reward gui…
Diffusion-Based Posterior Sampling: A Feynman-Kac Analysis of Bias and Stability
Matias G. Delgadino, Sebastien Motsch, Advait Parulekar +2
Diffusion-based posterior samplers use pretrained diffusion priors to sample from measurement- or reward-conditioned posteriors, and are widely used for inverse problems. Yet their…
Test-Time Anchoring for Discrete Diffusion Posterior Sampling
Litu Rout, Andreas Lugmayr, Yasamin Jafarian +4
While continuous diffusion models have achieved remarkable success, discrete diffusion offers a unified framework for jointly modeling text and images. Beyond unification, discrete…
Efficient Approximate Posterior Sampling with Annealed Langevin Monte Carlo
Advait Parulekar, Litu Rout, Karthikeyan Shanmugam +1
We study the problem of posterior sampling in the context of score based generative models. We have a trained score network for a prior , a measurement model , and ar…
Constrained Posterior Sampling: Time Series Generation with Hard Constraints
Sai Shankar Narasimhan, Shubhankar Agarwal, Litu Rout +2
Generating realistic time series samples is crucial for stress-testing models and protecting user privacy by using synthetic data. In engineering and safety-critical applications,…