Beyond the Dirac Delta: Mitigating Diversity Collapse in Reinforcement Fine-Tuning for Versatile Image Generation

Reinforcement learning (RL) has emerged as a powerful paradigm for fine-tuning large-scale generative models, such as diffusion and flow models, to align with complex human preferences and user-specified tasks. A fundamental limitation remains \textit{the curse of diversity collapse}, where the objective formulation and optimization landscape inherently collapse the policy to a Dirac delta distribution. To address this challenge, we propose \textbf{DRIFT} (\textbf{D}ive\textbf{R}sity-\textbf{I}ncentivized Reinforcement \textbf{F}ine-\textbf{T}uning for Versatile Image Generation), an innovative framework that systematically incentivizes output diversity throughout the on-policy fine-tuning process, reconciling strong task alignment with high generation diversity to enhance versatility essential for applications that demand diverse candidate generations. We approach the problem across three representative perspectives: i) \textbf{sampling} a reward-concentrated subset that filters out reward outliers to prevent premature collapse; ii) \textbf{prompting} with stochastic variations to expand the conditioning space, and iii) \textbf{optimization} of the intra-group diversity with a potential-based reward shaping mechanism. Experimental results show that DRIFT achieves superior Pareto dominance regarding task alignment and generation diversity, yielding a $ 9.08\%\!\sim\! 43.46\%$ increase in diversity at equivalent alignment levels and a $ 59.65\% \!\sim\! 65.86\%$ increase in alignment at equivalent levels of diversity.

Energy-Guided Flow Matching Enables Few-Step Conformer Generation and Ground-State Identification

Generating low-energy conformer ensembles and identifying ground-state conformations from molecular graphs remain computationally demanding with physics-based pipelines. Current learning-based approaches often suffer from a fragmented paradigm: generative models capture diversity but lack reliable energy calibration, whereas deterministic predictors target a single structure and fail to represent ensemble variability. Here we present EnFlow, a unified framework that couples flow matching (FM) with an explicitly learned energy model through an energy-guided sampling scheme defined along a non-Gaussian FM path. By incorporating energy-gradient guidance during sampling, our method steers trajectories toward lower-energy regions, substantially improving conformational fidelity, particularly in the few-step regime. The learned energy function further enables efficient energy-based ranking of generated ensembles for accurate ground-state identification. Extensive experiments on GEOM-QM9 and GEOM-Drugs demonstrate that EnFlow simultaneously improves generation metrics with 1--2 ODE-steps and reduces ground-state prediction errors compared with state-of-the-art methods.

Think Consistently, Reason Efficiently: Energy-Based Calibration for Implicit Chain-of-Thought

Large Language Models (LLMs) have demonstrated strong reasoning capabilities through \emph{Chain-of-Thought} (CoT) prompting, which enables step-by-step intermediate reasoning. However, explicit CoT methods rely on discrete token-level reasoning processes that are prone to error propagation and limited by vocabulary expressiveness, often resulting in rigid and inconsistent reasoning trajectories. Recent research has explored implicit or continuous reasoning in latent spaces, allowing models to perform internal reasoning before generating explicit output. Although such approaches alleviate some limitations of discrete CoT, they generally lack explicit mechanisms to enforce consistency among reasoning steps, leading to divergent reasoning paths and unstable outcomes. To address this issue, we propose EBM-CoT, an Energy-Based Chain-of-Thought Calibration framework that refines latent thought representations through an energy-based model (EBM). Our method dynamically adjusts latent reasoning trajectories toward lower-energy, high-consistency regions in the embedding space, improving both reasoning accuracy and consistency without modifying the base language model. Extensive experiments across mathematical, commonsense, and symbolic reasoning benchmarks demonstrate that the proposed framework significantly enhances the consistency and efficiency of multi-step reasoning in LLMs.

Right Question is Already Half the Answer: Fully Unsupervised LLM Reasoning Incentivization

While large language models (LLMs) have demonstrated exceptional capabilities in challenging tasks such as mathematical reasoning, existing methods to enhance reasoning ability predominantly rely on supervised fine-tuning (SFT) followed by reinforcement learning (RL) on reasoning-specific data after pre-training. However, these approaches critically depend on external supervisions--such as human labelled reasoning traces, verified golden answers, or pre-trained reward models--which limits scalability and practical applicability. In this work, we propose Entropy Minimized Policy Optimization (EMPO), which makes an early attempt at fully unsupervised LLM reasoning incentivization. EMPO does not require any supervised information for incentivizing reasoning capabilities (i.e., neither verifiable reasoning traces, problems with golden answers, nor additional pre-trained reward models). By continuously minimizing the predictive entropy of LLMs on unlabeled user queries in a latent semantic space, EMPO enables purely self-supervised evolution of reasoning capabilities with strong flexibility and practicality. Our experiments demonstrate competitive performance of EMPO on both mathematical reasoning and free-form commonsense reasoning tasks. Specifically, without any supervised signals, EMPO boosts the accuracy of Qwen2.5-Math-7B Base from 30.7\% to 48.1\% on mathematical benchmarks and improves truthfulness accuracy of Qwen2.5-7B Instruct from 87.16\% to 97.25\% on TruthfulQA.

InversionGNN: A Dual Path Network for Multi-Property Molecular Optimization

Exploring chemical space to find novel molecules that simultaneously satisfy multiple properties is crucial in drug discovery. However, existing methods often struggle with trading off multiple properties due to the conflicting or correlated nature of chemical properties. To tackle this issue, we introduce InversionGNN framework, an effective yet sample-efficient dual-path graph neural network (GNN) for multi-objective drug discovery. In the direct prediction path of InversionGNN, we train the model for multi-property prediction to acquire knowledge of the optimal combination of functional groups. Then the learned chemical knowledge helps the inversion generation path to generate molecules with required properties. In order to decode the complex knowledge of multiple properties in the inversion path, we propose a gradient-based Pareto search method to balance conflicting properties and generate Pareto optimal molecules. Additionally, InversionGNN is able to search the full Pareto front approximately in discrete chemical space. Comprehensive experimental evaluations show that InversionGNN is both effective and sample-efficient in various discrete multi-objective settings including drug discovery.

Measuring Diversity in Synthetic Datasets

Large language models (LLMs) are widely adopted to generate synthetic datasets for various natural language processing (NLP) tasks, such as text classification and summarization. However, accurately measuring the diversity of these synthetic datasets-an aspect crucial for robust model performance-remains a significant challenge. In this paper, we introduce DCScore, a novel method for measuring synthetic dataset diversity from a classification perspective. Specifically, DCScore formulates diversity evaluation as a sample classification task, leveraging mutual relationships among samples. We further provide theoretical verification of the diversity-related axioms satisfied by DCScore, highlighting its role as a principled diversity evaluation method. Experimental results on synthetic datasets reveal that DCScore enjoys a stronger correlation with multiple diversity pseudo-truths of evaluated datasets, underscoring its effectiveness. Moreover, both empirical and theoretical evidence demonstrate that DCScore substantially reduces computational costs compared to existing approaches. Code is available at: https://github.com/BlueWhaleLab/DCScore.

COME: Test-time adaption by Conservatively Minimizing Entropy

Machine learning models must continuously self-adjust themselves for novel data distribution in the open world. As the predominant principle, entropy minimization (EM) has been proven to be a simple yet effective cornerstone in existing test-time adaption (TTA) methods. While unfortunately its fatal limitation (i.e., overconfidence) tends to result in model collapse. For this issue, we propose to Conservatively Minimize the Entropy (COME), which is a simple drop-in replacement of traditional EM to elegantly address the limitation. In essence, COME explicitly models the uncertainty by characterizing a Dirichlet prior distribution over model predictions during TTA. By doing so, COME naturally regularizes the model to favor conservative confidence on unreliable samples. Theoretically, we provide a preliminary analysis to reveal the ability of COME in enhancing the optimization stability by introducing a data-adaptive lower bound on the entropy. Empirically, our method achieves state-of-the-art performance on commonly used benchmarks, showing significant improvements in terms of classification accuracy and uncertainty estimation under various settings including standard, life-long and open-world TTA, i.e., up to $34.5\%$ improvement on accuracy and $15.1\%$ on false positive rate.

The Best of Both Worlds: On the Dilemma of Out-of-distribution Detection

Out-of-distribution (OOD) detection is essential for model trustworthiness which aims to sensitively identify semantic OOD samples and robustly generalize for covariate-shifted OOD samples. However, we discover that the superior OOD detection performance of state-of-the-art methods is achieved by secretly sacrificing the OOD generalization ability. Specifically, the classification accuracy of these models could deteriorate dramatically when they encounter even minor noise. This phenomenon contradicts the goal of model trustworthiness and severely restricts their applicability in real-world scenarios. What is the hidden reason behind such a limitation? In this work, we theoretically demystify the ``\textit{sensitive-robust}'' dilemma that lies in many existing OOD detection methods. Consequently, a theory-inspired algorithm is induced to overcome such a dilemma. By decoupling the uncertainty learning objective from a Bayesian perspective, the conflict between OOD detection and OOD generalization is naturally harmonized and a dual-optimal performance could be expected. Empirical studies show that our method achieves superior performance on standard benchmarks. To our best knowledge, this work is the first principled OOD detection method that achieves state-of-the-art OOD detection performance without compromising OOD generalization ability. Our code is available at \href{https://github.com/QingyangZhang/DUL}{https://github.com/QingyangZhang/DUL}.

ConML: A Universal Meta-Learning Framework with Task-Level Contrastive Learning

Meta-learning enables learning systems to adapt quickly to new tasks, similar to humans. To emulate this human-like rapid learning and enhance alignment and discrimination abilities, we propose ConML, a universal meta-learning framework that can be applied to various meta-learning algorithms without relying on specific model architectures nor target models. The core of ConML is task-level contrastive learning, which extends contrastive learning from the representation space in unsupervised learning to the model space in meta-learning. By leveraging task identity as an additional supervision signal during meta-training, we contrast the outputs of the meta-learner in the model space, minimizing inner-task distance (between models trained on different subsets of the same task) and maximizing inter-task distance (between models from different tasks). We demonstrate that ConML integrates seamlessly with optimization-based, metric-based, and amortization-based meta-learning algorithms, as well as in-context learning, resulting in performance improvements across diverse few-shot learning tasks.

Probing the Safety Response Boundary of Large Language Models via Unsafe Decoding Path Generation

Large Language Models (LLMs) are implicit troublemakers. While they provide valuable insights and assist in problem-solving, they can also potentially serve as a resource for malicious activities. Implementing safety alignment could mitigate the risk of LLMs generating harmful responses. We argue that: even when an LLM appears to successfully block harmful queries, there may still be hidden vulnerabilities that could act as ticking time bombs. To identify these underlying weaknesses, we propose to use a cost value model as both a detector and an attacker. Trained on external or self-generated harmful datasets, the cost value model could successfully influence the original safe LLM to output toxic content in decoding process. For instance, LLaMA-2-chat 7B outputs 39.18% concrete toxic content, along with only 22.16% refusals without any harmful suffixes. These potential weaknesses can then be exploited via prompt optimization such as soft prompts on images. We name this decoding strategy: Jailbreak Value Decoding (JVD), emphasizing that seemingly secure LLMs may not be as safe as we initially believe. They could be used to gather harmful data or launch covert attacks.