Fine-Tuning Discrete Diffusion Models with Policy Gradient Methods

• URL: http://arxiv.org/abs/2502.01384v1
• Date: Mon, 03 Feb 2025 14:20:19 GMT
• Title: Fine-Tuning Discrete Diffusion Models with Policy Gradient Methods
• Authors: Oussama Zekri, Nicolas Boullé,
• Abstract summary: We propose an efficient, broadly applicable, and theoretically justified policy gradient algorithm for fine-tuning discrete diffusion models over non-differentiable rewards. Our numerical experiments across several discrete generative tasks demonstrate the scalability and efficiency of our method.
• Score: 4.028503203417233
• License:
• Abstract: Discrete diffusion models have recently gained significant attention due to their ability to process complex discrete structures for language modeling. However, fine-tuning these models with policy gradient methods, as is commonly done in Reinforcement Learning from Human Feedback (RLHF), remains a challenging task. We propose an efficient, broadly applicable, and theoretically justified policy gradient algorithm, called Score Entropy Policy Optimization (SEPO), for fine-tuning discrete diffusion models over non-differentiable rewards. Our numerical experiments across several discrete generative tasks demonstrate the scalability and efficiency of our method. Our code is available at https://github.com/ozekri/SEPO

Related papers

• Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics [7.873510219469276]
  We introduce two novel training methods for discrete diffusion samplers. These methods yield memory-efficient training and achieve state-of-the-art results in unsupervised optimization. We introduce adaptations of SN-NIS and Neural Chain Monte Carlo that enable for the first time the application of discrete diffusion models to this problem.
  arXiv  Detail & Related papers  (2025-02-12T18:59:55Z)
• Training-free Diffusion Model Alignment with Sampling Demons [15.400553977713914]
  We propose an optimization approach, dubbed Demon, to guide the denoising process at inference time without backpropagation through reward functions or model retraining. Our approach works by controlling noise distribution in denoising steps to concentrate density on regions corresponding to high rewards through optimization. To the best of our knowledge, the proposed approach is the first inference-time, backpropagation-free preference alignment method for diffusion models.
  arXiv  Detail & Related papers  (2024-10-08T07:33:49Z)
• Derivative-Free Guidance in Continuous and Discrete Diffusion Models with Soft Value-Based Decoding [84.3224556294803]
  Diffusion models excel at capturing the natural design spaces of images, molecules, DNA, RNA, and protein sequences. We aim to optimize downstream reward functions while preserving the naturalness of these design spaces. Our algorithm integrates soft value functions, which looks ahead to how intermediate noisy states lead to high rewards in the future.
  arXiv  Detail & Related papers  (2024-08-15T16:47:59Z)
• Understanding Reinforcement Learning-Based Fine-Tuning of Diffusion Models: A Tutorial and Review [63.31328039424469]
  This tutorial provides a comprehensive survey of methods for fine-tuning diffusion models to optimize downstream reward functions. We explain the application of various RL algorithms, including PPO, differentiable optimization, reward-weighted MLE, value-weighted sampling, and path consistency learning.
  arXiv  Detail & Related papers  (2024-07-18T17:35:32Z)
• Learning Multimodal Behaviors from Scratch with Diffusion Policy Gradient [26.675822002049372]
  Deep Diffusion Policy Gradient (DDiffPG) is a novel actor-critic algorithm that learns from scratch multimodal policies. DDiffPG forms a multimodal training batch and utilizes mode-specific Q-learning to mitigate the inherent greediness of the RL objective. Our approach further allows the policy to be conditioned on mode-specific embeddings to explicitly control the learned modes.
  arXiv  Detail & Related papers  (2024-06-02T09:32:28Z)
• Learning a Diffusion Model Policy from Rewards via Q-Score Matching [93.0191910132874]
  We present a theoretical framework linking the structure of diffusion model policies to a learned Q-function. We propose a new policy update method from this theory, which we denote Q-score matching.
  arXiv  Detail & Related papers  (2023-12-18T23:31:01Z)
• When to Update Your Model: Constrained Model-based Reinforcement Learning [50.74369835934703]
  We propose a novel and general theoretical scheme for a non-decreasing performance guarantee of model-based RL (MBRL) Our follow-up derived bounds reveal the relationship between model shifts and performance improvement. A further example demonstrates that learning models from a dynamically-varying number of explorations benefit the eventual returns.
  arXiv  Detail & Related papers  (2022-10-15T17:57:43Z)
• Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learning [70.20191211010847]
  Offline reinforcement learning (RL) aims to learn an optimal policy using a previously collected static dataset. We introduce Diffusion Q-learning (Diffusion-QL) that utilizes a conditional diffusion model to represent the policy. We show that our method can achieve state-of-the-art performance on the majority of the D4RL benchmark tasks.
  arXiv  Detail & Related papers  (2022-08-12T09:54:11Z)
• Model-Augmented Actor-Critic: Backpropagating through Paths [81.86992776864729]
  Current model-based reinforcement learning approaches use the model simply as a learned black-box simulator. We show how to make more effective use of the model by exploiting its differentiability.
  arXiv  Detail & Related papers  (2020-05-16T19:18:10Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.