Any well-behaved generative model over a variable $\mathbf{x}$ can be expressed as a deterministic transformation of an exogenous (outsourced’) Gaussian noise variable $\mathbf{z}$: $\mathbf{x}=f_\theta(\mathbf{z})$. In such a model (eg, a VAE, GAN, or continuous-time flow-based model), sampling of the target variable $\mathbf{x} \sim p_\theta(\mathbf{x})$ is straightforward, but sampling from a posterior distribution of the form $p(\mathbf{x}\mid\mathbf{y}) \propto p_\theta(\mathbf{x})r(\mathbf{x},\mathbf{y})$, where $r$ is a constraint function depending on an auxiliary variable $\mathbf{y}$, is generally intractable. We propose to amortize the cost of sampling from such posterior distributions with diffusion models that sample a distribution in the noise space ($\mathbf{z}$). These diffusion samplers are trained by reinforcement learning algorithms to enforce that the transformed samples $f_\theta(\mathbf{z})$ are distributed according to the posterior in the data space ($\mathbf{x}$). For many models and constraints, the posterior in noise space is smoother than in data space, making it more suitable for amortized inference. Our method enables conditional sampling under unconditional GAN, (H)VAE, and flow-based priors, comparing favorably with other inference methods. We demonstrate the proposed outsourced diffusion sampling in several experiments with large pretrained prior models: conditional image generation, reinforcement learning with human feedback, and protein structure generation.

Reinforcement learning (RL) is a critical component of large language model (LLM) post-training. However, on-policy algorithms used for post-training are not naturally robust to a diversified content of experience replay buffers, which asynchronous off-policy actors can efficiently populate in parallel to training. We propose efficiently learning on such off-policy data via Trajectory Balance with Asynchrony (TBA), an approach to asynchronous RL for LLMs that leverages the principled off-policy TB objective. On math, preference-tuning, and automated red-teaming tasks, we post-train models ranging from Pythia 410M to Qwen 2.5 7B, finding TBA offers speed and performance boosts over strong baselines like Online DPO and Dr. GRPO. Beyond TBA's performance benefits (high accuracy even as asynchrony grows) and speedups ($4\times$ or more), we show its reward- and recency-prioritizing sampling enable further gains as data generation is scaled. Our code is available at https://github.com/bbartoldson/TBA.

Offline reinforcement learning (RL) holds promise as a means to learn high-reward policies from a static dataset, without the need for further environment interactions. However, a key challenge in offline RL lies in effectively stitching portions of suboptimal trajectories from the static dataset while avoiding extrapolation errors arising due to a lack of support in the dataset. Existing approaches use conservative methods that are tricky to tune and struggle with multi-modal data or rely on noisy Monte Carlo return-to-go samples for reward conditioning. In this work, we propose a novel approach that leverages the expressiveness of latent diffusion to model in-support trajectory sequences as compressed latent skills. This facilitates learning a Q-function while avoiding extrapolation error via batch-constraining. The latent space is also expressive and gracefully copes with multi-modal data. We show that the learned temporally-abstract latent space encodes richer task-specific information for offline RL tasks as compared to raw state-actions. This improves credit assignment and facilitates faster reward propagation during Q-learning. Our method demonstrates state-of-the-art performance on the D4RL benchmarks, particularly excelling in long-horizon, sparse-reward tasks.

Diffusion models have emerged as effective distribution estimators in vision, language, and reinforcement learning, but their use as priors in downstream tasks poses an intractable posterior inference problem. This paper studies *amortized* sampling of the posterior over data, $\mathbf{x}\sim p^{\rm post}(\mathbf{x})\propto p(\mathbf{x})r(\mathbf{x})$, in a model that consists of a diffusion generative model prior $p(\mathbf{x})$ and a black-box constraint or likelihood function $r(\mathbf{x})$. We state and prove the asymptotic correctness of a data-free learning objective, *relative trajectory balance*, for training a diffusion model that samples from this posterior, a problem that existing methods solve only approximately or in restricted cases. Relative trajectory balance arises from the generative flow network perspective on diffusion models, which allows the use of deep reinforcement learning techniques to improve mode coverage. Experiments illustrate the broad potential of unbiased inference of arbitrary posteriors under diffusion priors: in vision (classifier guidance), language (infilling under a discrete diffusion LLM), and multimodal data (text-to-image generation). Beyond generative modeling, we apply relative trajectory balance to the problem of continuous control with a score-based behavior prior, achieving state-of-the-art results on benchmarks in offline reinforcement learning. Code is available at [this link](https://github.com/GFNOrg/diffusion-finetuning).

Agents that can build temporally abstract representations of their environment are better able to understand their world and make plans on extended time scales, with limited computational power and modeling capacity. However, existing methods for automatically learning temporally abstract world models usually require millions of online environmental interactions and incentivize agents to reach every accessible environmental state, which is infeasible for most real-world robots both in terms of data efficiency and hardware safety. In this paper, we present an approach for simultaneously learning sets of skills and temporally abstract, skill-conditioned world models purely from offline data, enabling agents to perform zero-shot online planning of skill sequences for new tasks. We show that our approach performs comparably to or better than a wide array of state-of-the-art offline RL algorithms on a number of simulated robotics locomotion and manipulation benchmarks, while offering a higher degree of adaptability to new goals. Finally, we show that our approach offers a much higher degree of robustness to perturbations in environmental dynamics, compared to policy-based methods.

The successful application of robotic control requires intelligent decision-making to handle the long tail of complex scenarios that arise in real-world environments. Recently, Deep Reinforcement Learning (DRL) has provided a data-driven framework to automatically learn effective policies in such complex settings. Since its introduction in 2018, Soft Actor-Critic (SAC) remains as one of the most popular off-policy DRL algorithms and has been used extensively to learn performant robotic control policies. However, in this paper we argue that by relying on the maximum entropy formalism to define learning objectives, previous work introduces a significant bias away from optimal decision making, which often requires near-deterministic behaviour for high-precision tasks. Moreover, we show that when training with the original variants of SAC, overcoming this bias by reducing entropy budgets or entropy coefficients introduces separate issues that lead to slow or unstable learning. We address these shortcomings by treating the entropy coefficient$\alpha$as a random variable and introduce Multi-Alpha Soft Actor-Critic (MAS). We show how MAS overcomes the stochastic bias of SAC in a variety of robotic control tasks including the CARLA urban-driving simulator, while maintaining the stability and sample efficiency of the original algorithms.