Aligning diffusion models with downstream objectives is essential for their practical applications. However, standard alignment methods often struggle with step generalization when directly applied to few-step diffusion models, leading to inconsistent performance across different denoising step scenarios. To address this, we introduce Stepwise Diffusion Policy Optimization (SDPO), a novel alignment method tailored for few-step diffusion models. Unlike prior approaches that rely on a single sparse reward from only the final step of each denoising trajectory for trajectory-level optimization, SDPO incorporates dense reward feedback at every intermediate step. By learning the differences in dense rewards between paired samples, SDPO facilitates stepwise optimization of few-step diffusion models, ensuring consistent alignment across all denoising steps. To promote stable and efficient training, SDPO introduces an online reinforcement learning framework featuring several novel strategies designed to effectively exploit the stepwise granularity of dense rewards. Experimental results demonstrate that SDPO consistently outperforms prior methods in reward-based alignment across diverse step configurations, underscoring its robust step generalization capabilities. Code is avaliable at https://github.com/ZiyiZhang27/sdpo.

Multimodal Large Language Model (MLLM) have demonstrated strong generalization capabilities across diverse distributions and tasks, largely due to extensive pre-training datasets. Fine-tuning MLLM has become a common practice to improve performance on specific downstream tasks. However, during fine-tuning, MLLM often faces the risk of forgetting knowledge acquired during pre-training, which can result in a decline in generalization abilities. To balance the trade-off between generalization and specialization, we propose measuring the parameter importance for both pre-trained and fine-tuning distributions, based on frozen pre-trained weight magnitude and accumulated fine-tuning gradient values. We further apply an importance-aware weight allocation strategy, selectively updating relatively important parameters for downstream tasks. We conduct empirical evaluations on both image captioning and visual question-answering tasks using various MLLM architectures. The comprehensive experimental analysis demonstrates the effectiveness of the proposed solution, highlighting the efficiency of the crucial modules in enhancing downstream specialization performance while mitigating generalization degradation in MLLM Fine-Tuning.

Minimax optimization is gaining increasing attention in modern machine learning applications. Driven by large-scale models and massive volumes of data collected from edge devices, as well as the concern to preserve client privacy, communication-efficient distributed minimax optimization algorithms become popular, such as Local Stochastic Gradient Descent Ascent (Local-SGDA), and Local Decentralized SGDA (Local-DSGDA). While most existing research on distributed minimax algorithms focuses on convergence rates, computation complexity, and communication efficiency, the generalization performance remains underdeveloped, whereas generalization ability is a pivotal indicator for evaluating the holistic performance of a model when fed with unknown data. In this paper, we propose the stability-based generalization analytical framework for Distributed-SGDA, which unifies two popular distributed minimax algorithms including Local-SGDA and Local-DSGDA, and conduct a comprehensive analysis of stability error, generalization gap, and population risk across different metrics under various settings, e.g., (S)C-(S)C, PL-SC, and NC-NC cases. Our theoretical results reveal the trade-off between the generalization gap and optimization error and suggest hyperparameters choice to obtain the optimal population risk.

Offline reinforcement learning (RL) methods harness previous experiences to derive an optimal policy, forming the foundation for pre-trained large-scale models (PLMs). When encountering tasks not seen before, PLMs often utilize several expert trajectories as prompts to expedite their adaptation to new requirements. Though a range of prompt-tuning methods have been proposed to enhance the quality of prompts, these methods often face optimization restrictions due to prompt initialization, which can significantly constrain the exploration domain and potentially lead to suboptimal solutions. To eliminate the reliance on the initial prompt, we shift our perspective towards the generative model, framing the prompt-tuning process as a form of conditional generative modeling, where prompts are generated from random noise. Our innovation, the Prompt Diffuser, leverages a conditional diffusion model to produce prompts of exceptional quality. Central to our framework is the approach to trajectory reconstruction and the meticulous integration of downstream task guidance during the training phase. Further experimental results underscore the potency of the Prompt Diffuser as a robust and effective tool for the prompt-tuning process, demonstrating strong performance in the meta-RL tasks.

The purpose of offline multi-task reinforcement learning (MTRL) is to develop a unified policy applicable to diverse tasks without the need for online environmental interaction. Recent advancements approach this through sequence modeling, leveraging the Transformer architecture's scalability and the benefits of parameter sharing to exploit task similarities. However, variations in task content and complexity pose significant challenges in policy formulation, necessitating judicious parameter sharing and management of conflicting gradients for optimal policy performance. Furthermore, identifying the optimal parameter subspace for each task often necessitates prior knowledge of the task identifier during inference, limiting applicability in real-world scenarios with variable task content and unknown current tasks. In this work, we introduce the Harmony Multi-Task Decision Transformer (HarmoDT), a novel solution designed to identify an optimal harmony subspace of parameters for each task. We formulate this as a bi-level optimization problem within a meta-learning framework, where the upper level learns masks to define the harmony subspace, while the inner level focuses on updating parameters to improve the overall performance of the unified policy. To eliminate the need for task identifiers, we further design a group-wise variant (G-HarmoDT) that clusters tasks into coherent groups based on gradient information, and utilizes a gating network to determine task identifiers during inference. Empirical evaluations across various benchmarks highlight the superiority of our approach, demonstrating its effectiveness in the multi-task context with specific improvements of 8% gain in task-provided settings, 5% in task-agnostic settings, and 10% in unseen settings.

In numerous artificial intelligence applications, the collaborative efforts of multiple intelligent agents are imperative for the successful attainment of target objectives. To enhance coordination among these agents, a distributed communication framework is often employed. However, indiscriminate information sharing among all agents can be resource-intensive, and the adoption of manually pre-defined communication architectures imposes constraints on inter-agent communication, thus limiting the potential for effective collaboration. Moreover, the communication framework often remains static during inference, which may result in sustained high resource consumption, as in most cases, only key decisions necessitate information sharing among agents. In this study, we introduce a novel approach wherein we conceptualize the communication architecture among agents as a learnable graph. We formulate this problem as the task of determining the communication graph while enabling the architecture parameters to update normally, thus necessitating a bi-level optimization process. Utilizing continuous relaxation of the graph representation and incorporating attention units, our proposed approach, CommFormer, efficiently optimizes the communication graph and concurrently refines architectural parameters through gradient descent in an end-to-end manner. Additionally, we introduce a temporal gating mechanism for each agent, enabling dynamic decisions on whether to receive shared information at a given time, based on current observations, thus improving decision-making efficiency. Extensive experiments on a variety of cooperative tasks substantiate the robustness of our model across diverse cooperative scenarios, where agents are able to develop more coordinated and sophisticated strategies regardless of changes in the number of agents.

Massive reinforcement learning (RL) data are typically collected to train policies offline without the need for interactions, but the large data volume can cause training inefficiencies. To tackle this issue, we formulate offline behavior distillation (OBD), which synthesizes limited expert behavioral data from sub-optimal RL data, enabling rapid policy learning. We propose two naive OBD objectives, DBC and PBC, which measure distillation performance via the decision difference between policies trained on distilled data and either offline data or a near-expert policy. Due to intractable bi-level optimization, the OBD objective is difficult to minimize to small values, which deteriorates PBC by its distillation performance guarantee with quadratic discount complexity $\mathcal{O}(1/(1-\gamma)^2)$. We theoretically establish the equivalence between the policy performance and action-value weighted decision difference, and introduce action-value weighted PBC (Av-PBC) as a more effective OBD objective. By optimizing the weighted decision difference, Av-PBC achieves a superior distillation guarantee with linear discount complexity $\mathcal{O}(1/(1-\gamma))$. Extensive experiments on multiple D4RL datasets reveal that Av-PBC offers significant improvements in OBD performance, fast distillation convergence speed, and robust cross-architecture/optimizer generalization.

Hallucinations in Large Language Models (LLMs) remain a major obstacle, particularly in high-stakes applications where factual accuracy is critical. While representation editing and reading methods have made strides in reducing hallucinations, their heavy reliance on specialised tools and training on in-domain samples, makes them difficult to scale and prone to overfitting. This limits their accuracy gains and generalizability to diverse datasets. This paper presents a lightweight method, Norm Voting (NoVo), which harnesses the untapped potential of attention head norms to dramatically enhance factual accuracy in zero-shot multiple-choice questions (MCQs). NoVo begins by automatically selecting truth-correlated head norms with an efficient, inference-only algorithm using only 30 random samples, allowing NoVo to effortlessly scale to diverse datasets. Afterwards, selected head norms are employed in a simple voting algorithm, which yields significant gains in prediction accuracy. NoVo demonstrates exceptional generalization to 20 diverse datasets, with significant gains in over 90% of them, far exceeding all current representation editing and reading methods. NoVo also reveals promising gains to finetuning strategies and building textual adversarial defence. NoVo's effectiveness with head norms opens new frontiers in LLM interpretability, robustness and reliability. One of the most significant challenges facing Large Language Models (LLMs) today is their tendency to hallucinate--outputs that are factually incorrect or entirely fabricated (Zhang et al., 2023b). This flaw is particularly serious in high-stakes applications like finance and healthcare, where even small errors can lead to huge losses and compromised patient safety (Kang & Liu, 2023; Pal et al., 2023). Reducing factual hallucinations is a critical research area with major practical benefits, essential for realising the full potential of LLMs to revolutionise these industries by enhancing efficiency and decision-making, and safeguarding against costly and harmful errors (Kaddour et al., 2023). Given these serious risks and the high cost of retraining LLMs, it is crucial to find affordable techniques to reduce factual hallucinations. Although inference techniques such as retrieval augmentation and prompt engineering work well, they come with significant limitations: latency and external dependencies, and the need for user expertise, respectively (Zhao et al., 2024; Sahoo et al., 2024). In response, we turn to representation editing and reading methods (REAR) (Zou et al., 2023), which operate within the model, ensuring rapid response times and eliminating the need for external data or user interaction.

Multi-task learning (MTL) leverages a shared model to accomplish multiple tasks and facilitate knowledge transfer. Recent research on task arithmetic-based MTL demonstrates that merging the parameters of independently fine-tuned models can effectively achieve MTL. However, existing merging methods primarily seek a static optimal solution within the original model parameter space, which often results in performance degradation due to the inherent diversity among tasks and potential interferences. To address this challenge, in this paper, we propose a Weight-Ensembling Mixture of Experts (WEMoE) method for multi-task model merging. Specifically, we first identify critical (or sensitive) modules by analyzing parameter variations in core modules of Transformer-based models before and after finetuning. Then, our WEMoE statically merges non-critical modules while transforming critical modules into a mixture-of-experts (MoE) structure. During inference, expert modules in the MoE are dynamically merged based on input samples, enabling a more flexible and adaptive merging approach. Building on WEMoE, we further introduce an efficient-and-effective WEMoE (E-WEMoE) method, whose core mechanism involves eliminating non-essential elements in the critical modules of WEMoE and implementing shared routing across multiple MoE modules, thereby significantly reducing both the trainable parameters, the overall parameter count, and computational overhead of the merged model by WEMoE. Experimental results across various architectures and tasks demonstrate that both WEMoE and E-WEMoE outperform state-of-the-art (SOTA) model merging methods in terms of MTL performance, generalization, and robustness.

Continual offline reinforcement learning (CORL) has shown impressive ability in diffusion-based lifelong learning systems by modeling the joint distributions of trajectories. However, most research only focuses on limited continual task settings where the tasks have the same observation and action space, which deviates from the realistic demands of training agents in various environments. In view of this, we propose Vector-Quantized Continual Diffuser, named VQ-CD, to break the barrier of different spaces between various tasks. Specifically, our method contains two complementary sections, where the quantization spaces alignment provides a unified basis for the selective weights activation. In the quantized spaces alignment, we leverage vector quantization to align the different state and action spaces of various tasks, facilitating continual training in the same space. Then, we propose to leverage a unified diffusion model attached by the inverse dynamic model to master all tasks by selectively activating different weights according to the task-related sparse masks. Finally, we conduct extensive experiments on 15 continual learning (CL) tasks, including conventional CL task settings (identical state and action spaces) and general CL task settings (various state and action spaces). Compared with 16 baselines, our method reaches the SOTA performance.