The rapid advancement of LLMs sparked significant interest in their potential to augment or automate managerial functions. One of the most recent trends in AI benchmarking is performance of Large Language Models (LLMs) over longer time horizons. While LLMs excel at tasks involving natural language and pattern recognition, their capabilities in multi-step, strategic business decision-making remain largely unexplored. Few studies demonstrated how results can be different from benchmarks in short-term tasks, as Vending-Bench revealed. Meanwhile, there is a shortage of alternative benchmarks for long-term coherence. This research analyses a novel benchmark using a business game for the decision making in business. The research contributes to the recent literature on AI by proposing a reproducible, open-access management simulator to the research community for LLM benchmarking. This novel framework is used for evaluating the performance of five leading LLMs available in free online interface: Gemini, ChatGPT, Meta AI, Mistral AI, and Grok. LLM makes decisions for a simulated retail company. A dynamic, month-by-month management simulation provides transparently in spreadsheet model as experimental environment. In each of twelve months, the LLMs are provided with a structured prompt containing a full business report from the previous period and are tasked with making key strategic decisions: pricing, order size, marketing budget, hiring, dismissal, loans, training expense, R&D expense, sales forecast, income forecast The methodology is designed to compare the LLMs on quantitative metrics: profit, revenue, and market share, and other KPIs. LLM decisions are analyzed in their strategic coherence, adaptability to market changes, and the rationale provided for their decisions. This approach allows to move beyond simple performance metrics for assessment of the long-term decision-making.

Large audio-language models are advancing rapidly, yet most evaluations emphasize speech or globally sourced sounds, overlooking culturally distinctive cues. This gap raises a critical question: can current models generalize to localized, non-semantic audio that communities instantly recognize but outsiders do not? To address this, we present TAU (Taiwan Audio Understanding), a benchmark of everyday Taiwanese "soundmarks." TAU is built through a pipeline combining curated sources, human editing, and LLM-assisted question generation, producing 702 clips and 1,794 multiple-choice items that cannot be solved by transcripts alone. Experiments show that state-of-the-art LALMs, including Gemini 2.5 and Qwen2-Audio, perform far below local humans. TAU demonstrates the need for localized benchmarks to reveal cultural blind spots, guide more equitable multimodal evaluation, and ensure models serve communities beyond the global mainstream.

Autoregressive (AR) large language models (LLMs) have achieved remarkable performance across a wide range of natural language tasks, yet their inherent sequential decoding limits inference efficiency. In this work, we propose Fast-dLLM v2, a carefully designed block diffusion language model (dLLM) that efficiently adapts pretrained AR models into dLLMs for parallel text generation, requiring only approximately 1B tokens of fine-tuning. This represents a 500x reduction in training data compared to full-attention diffusion LLMs such as Dream (580B tokens), while preserving the original model's performance. Our approach introduces a novel training recipe that combines a block diffusion mechanism with a complementary attention mask, enabling blockwise bidirectional context modeling without sacrificing AR training objectives. To further accelerate decoding, we design a hierarchical caching mechanism: a block-level cache that stores historical context representations across blocks, and a sub-block cache that enables efficient parallel generation within partially decoded blocks. Coupled with our parallel decoding pipeline, Fast-dLLM v2 achieves up to 2.5x speedup over standard AR decoding without compromising generation quality. Extensive experiments across diverse benchmarks demonstrate that Fast-dLLM v2 matches or surpasses AR baselines in accuracy, while delivering state-of-the-art efficiency among dLLMs - marking a significant step toward the practical deployment of fast and accurate LLMs. Code and model will be publicly released.

Supervised fine-tuning (SFT) is the predominant method for adapting large language models (LLMs), yet it often struggles with generalization compared to reinforcement learning (RL). In this work, we posit that this performance disparity stems not just from the loss function, but from a more fundamental difference: SFT learns from a fixed, pre-collected dataset, whereas RL utilizes on-policy data sampled from the current policy. Building on this hypothesis, we introduce one-token rollout (OTR), a novel fine-tuning algorithm that guides SFT with the policy gradient method. At each step, it performs a Monte Carlo "rollout" by sampling multiple candidate tokens from the current policy's distribution. The ground-truth token from the supervised data is then used to provide a reward signal to these samples. Through extensive experiments on a diverse suite of challenging benchmarks spanning mathematical reasoning, code generation, and general domain reasoning, we demonstrate that OTR consistently outperforms standard SFT. Our findings establish OTR as a powerful and practical alternative for fine-tuning LLMs and provide compelling evidence that the on-policy nature of data is a critical driver of generalization, offering a promising new direction for fine-tuning LLMs. Supervised fine-tuning (SFT) has become a cornerstone for adapting large language models (LLMs) to downstream tasks (Ouyang et al., 2022; Chung et al., 2022; Zhang et al., 2025).

The capacity of Large Language Models (LLMs) to follow complex instructions and generate factually accurate text is critical for their real-world application. However, standard decoding methods often fail to robustly satisfy these requirements, while existing control techniques frequently degrade general output quality. In this work, we introduce Attribution-Guided Decoding (AGD), an interpretability-based decoding strategy. Instead of directly manipulating model activations, AGD considers a set of high-probability output token candidates and selects the one that exhibits the highest attribution to a user-defined Region of Interest (ROI). This ROI can be flexibly defined over different parts of the model's input or internal components, allowing AGD to steer generation towards various desirable behaviors. We demonstrate AGD's efficacy across three challenging domains. For instruction following, we show that AGD significantly boosts adherence (e.g., improving the overall success rate on Llama 3.1 from 66.0% to 79.1%). For knowledge-intensive tasks, we show that guiding generation towards usage of internal knowledge components or contextual sources can reduce hallucinations and improve factual accuracy in both closed-book and open-book settings. Furthermore, we propose an adaptive, entropy-based variant of AGD that mitigates quality degradation and reduces computational overhead by applying guidance only when the model is uncertain. Our work presents a versatile, more interpretable, and effective method for enhancing the reliability of modern LLMs. Large Language Models (LLMs) have emerged as powerful tools capable of generating fluent, coherent and contextually relevant text across numerous applications (Brown et al., 2020; Ouyang et al., 2022; Achiam et al., 2023). Despite their success, their reliability is undermined by critical failures, most notably inconsistent adherence to user instructions (Sun et al., 2023; Zhou et al., 2023a; Zeng et al., 2024) and a tendency to generate non-factual, or hallucinated (Wei et al., 2024), information.

Dialogue summarization aims to distill the core meaning of a conversation into a concise text. This is crucial for reducing the complexity and noise inherent in dialogue-heavy applications. While recent approaches typically train language models to mimic human-written summaries, such supervision is costly and often results in outputs that lack task-specific focus limiting their effectiveness in downstream applications, such as medical tasks. In this paper, we propose \app, a framework for task-oriented utility-based dialogue summarization. \app starts by generating multiple summaries and task-oriented question-answer pairs from a dialogue in a zero-shot manner using a pool of large language models (LLMs). The quality of the generated summaries is evaluated by having LLMs answer task-related questions before \textit{(i)} selecting the best candidate answers and \textit{(ii)} identifying the most informative summary based on these answers. Finally, we fine-tune the best LLM on the selected summaries. When validated on multiple datasets, \app demonstrates its effectiveness by achieving competitive results in various zero-shot settings, rivaling fully-supervised State-of-the-Art (SotA) methods.

We introduce Flower, a solver for inverse problems. It leverages a pre-trained flow model to produce reconstructions that are consistent with the observed measurements. Flower operates through an iterative procedure over three steps: (i) a flow-consistent destination estimation, where the velocity network predicts a denoised target; (ii) a refinement step that projects the estimated destination onto a feasible set defined by the forward operator; and (iii) a time-progression step that re-projects the refined destination along the flow trajectory. We provide a theoretical analysis that demonstrates how Flower approximates Bayesian posterior sampling, thereby unifying perspectives from plug-and-play methods and generative inverse solvers. On the practical side, Flower achieves state-of-the-art reconstruction quality while using nearly identical hyperparameters across various inverse problems.

The efficient distributed training of Large Language Models (LLMs) is severely hampered by the extreme variance in context lengths. This data heterogeneity, amplified by conventional packing strategies and asymmetric forward-backward costs, leads to critical inefficiencies such as cascading workload imbalances and severe hardware underutilization. Existing solutions attempt to mitigate these challenges, but often at the expense of memory or communication efficiency. To address these challenges, we introduce SlimPack, a framework that fundamentally rethinks data packing and scheduling by decomposing samples into fine-grained slices. This slice-level decomposition immediately mitigates critical memory and communication bottlenecks by transforming large, volatile workloads into a stream of smaller, manageable units. This flexibility is then harnessed for our core innovation, Asymmetric Partitioning, which assembles balanced scheduling units uniquely optimized for the different demands of the forward and backward passes. Orchestrated by a two-phase solver and a high-fidelity simulator, SlimPack holistically resolves imbalances across all parallel dimensions. Extensive experiments demonstrate that SlimPack achieves up to a $2.8\times$ training throughput improvement over baselines, breaking the conventional trade-off by delivering both superior balance and high resource efficiency.

Creating personalized 3D animations with precise control and realistic head motions remains challenging for current speech-driven 3D facial animation methods. Editing these animations is especially complex and time consuming, requires precise control and typically handled by highly skilled animators. Most existing works focus on controlling style or emotion of the synthesized animation and cannot edit/regenerate parts of an input animation. They also overlook the fact that multiple plausible lip and head movements can match the same audio input. T o address these challenges, we present 3DiF ACE, a novel method for holistic speech-driven 3D facial animation. Our approach produces diverse plausible lip and head motions for a single audio input and allows for editing via keyframing and interpolation. Specifically, we propose a fully-convolutional diffusion model that can leverage the viseme-level diversity in our training corpus. Additionally, we employ a speaking-style personalization and a novel sparsely-guided motion diffusion to enable precise control and editing. Through quantitative and qualitative evaluations, we demonstrate that our method is capable of generating and editing diverse holistic 3D facial animations given a single audio input, with control between high fidelity and diversity.

ABSTRACT Transformer-based models have become the state of the art across multiple domains, from natural language processing to machine listening, thanks to attention mechanisms. However, the attention layers require a large number of parameters and high-end hardware for both training and inference. We propose a novel pruning technique targeted explicitly at the attention mechanism, where we decouple the pruning of the four layers in the attention block, namely: query, keys, values and outputs' projection matrices. We also investigate pruning strategies to prune along the head and channel dimensions, and compare the performance of the Audio Spectrogram Transformer (AST) [1] model under different pruning scenarios. Our results show that even by pruning 50% of the attention parameters we incur in performance degradation of less than 1%.