Transformers replace recurrence with a memory that grows with sequence length and self-attention that enables ad-hoc look ups over past tokens. Consequently, they lack an inherent incentive to compress history into compact latent states with consistent transition rules. This often leads to learning solutions that generalize poorly. We introduce Next-Latent Prediction (NextLat), which extends standard next-token training with self-supervised predictions in the latent space. Specifically, NextLat trains a transformer to learn latent representations that are predictive of its next latent state given the next output token. Theoretically, we show that these latents provably converge to belief states, compressed information of the history necessary to predict the future. This simple auxiliary objective also injects a recurrent inductive bias into transformers, while leaving their architecture, parallel training, and inference unchanged. NextLat effectively encourages the transformer to form compact internal world models with its own belief states and transition dynamics -- a crucial property absent in standard next-token prediction transformers. Empirically, across benchmarks targeting core sequence modeling competencies -- world modeling, reasoning, planning, and language modeling -- NextLat demonstrates significant gains over standard next-token training in downstream accuracy, representation compression, and lookahead planning. NextLat stands as a simple and efficient paradigm for shaping transformer representations toward stronger generalization.

Patients with chronic obstructive pulmonary disease (COPD) have an increased risk of hospitalizations, strongly associated with decreased survival, yet predicting the timing of these events remains challenging and has received limited attention in the literature. In this study, we performed survival analysis to predict hospitalization and death in COPD patients using longitudinal electronic health records (EHRs), comparing statistical models, machine learning (ML), and deep learning (DL) approaches. We analyzed data from more than 150k patients from the SIDIAP database in Catalonia, Spain, from 2013 to 2017, modeling hospitalization as a first event and death as a semi-competing terminal event. Multiple models were evaluated, including Cox proportional hazards, SurvivalBoost, DeepPseudo, SurvTRACE, Dynamic Deep-Hit, and Deep Recurrent Survival Machine. Results showed that DL models utilizing recurrent architectures outperformed both ML and linear approaches in concordance and time-dependent AUC, especially for hospitalization, which proved to be the harder event to predict. This study is, to our knowledge, the first to apply deep survival analysis on longitudinal EHR data to jointly predict multiple time-to-event outcomes in COPD patients, highlighting the potential of DL approaches to capture temporal patterns and improve risk stratification.

Despite the proliferation of powerful agentic models, the lack of critical post-training details hinders the development of strong counterparts in the open-source community. In this study, we present a comprehensive and fully open-source pipeline for training a high-performance agentic model for interacting with external tools and environments, named Klear-Qwen3-AgentForge, starting from the Qwen3-8B base model. We design effective supervised fine-tuning (SFT) with synthetic data followed by multi-turn reinforcement learning (RL) to unlock the potential for multiple diverse agentic tasks. We perform exclusive experiments on various agentic benchmarks in both tool use and coding domains. Klear-Qwen3-AgentForge-8B achieves state-of-the-art performance among LLMs of similar size and remains competitive with significantly larger models.

Due to their ability of follow natural language instructions, vision-language-action (VLA) models are increasingly prevalent in the embodied AI arena, following the widespread success of their precursors -- LLMs and VLMs. In this paper, we discuss 10 principal milestones in the ongoing development of VLA models -- multimodality, reasoning, data, evaluation, cross-robot action generalization, efficiency, whole-body coordination, safety, agents, and coordination with humans. Furthermore, we discuss the emerging trends of using spatial understanding, modeling world dynamics, post training, and data synthesis -- all aiming to reach these milestones. Through these discussions, we hope to bring attention to the research avenues that may accelerate the development of VLA models into wider acceptability.

Reinforcement learning (RL) is often credited with improving language model reasoning and generalization at the expense of degrading memorized knowledge. We challenge this narrative by observing that RL-enhanced models consistently outperform their base and supervised fine-tuned (SFT) counterparts on pure knowledge recall tasks, particularly those requiring traversal of hierarchical, structured knowledge (e.g., medical codes). We hypothesize these gains stem not from newly acquired data, but from improved procedural skills in navigating and searching existing knowledge hierarchies within the model parameters. To support this hypothesis, we show that structured prompting, which explicitly guides SFTed models through hierarchical traversal, recovers most of the performance gap (reducing 24pp to 7pp on MedConceptsQA for DeepSeek-V3/R1). We further find that while prompting improves final-answer accuracy, RL-enhanced models retain superior ability to recall correct procedural paths on deep-retrieval tasks. Finally our layer-wise internal activation analysis reveals that while factual representations (e.g., activations for the statement "code 57.95 refers to urinary infection") maintain high cosine similarity between SFT and RL models, query representations (e.g., "what is code 57.95") diverge noticeably, indicating that RL primarily transforms how models traverse knowledge rather than the knowledge representation itself.

We introduce a unified attention-based framework for joint score and density estimation. Framing the problem as a sequence-to-sequence task, we develop a permutation- and affine-equivariant transformer that estimates both the probability density $f(x)$ and its score $\nabla_x \log f(x)$ directly from i.i.d. samples. Unlike traditional score-matching methods that require training a separate model for each distribution, our approach learns a single distribution-agnostic operator that generalizes across densities and sample sizes. The architecture employs cross-attention to connect observed samples with arbitrary query points, enabling generalization beyond the training data, while built-in symmetry constraints ensure equivariance to permutation and affine transformations. Analytically, we show that the attention weights can recover classical kernel density estimation (KDE), and verify it empirically, establishing a principled link between classical KDE and the transformer architecture. Empirically, the model achieves substantially lower error and better scaling than KDE and score-debiased KDE (SD-KDE), while exhibiting better runtime scaling. Together, these results establish transformers as general-purpose, data-adaptive operators for nonparametric density and score estimation.

Voice-controlled dialog systems have become immensely popular due to their ability to perform a wide range of actions in response to diverse user queries. These agents possess a predefined set of skills or intents to fulfill specific user tasks. But every system has its own limitations. There are instances where, even for known intents, if any model exhibits low confidence, it results in rejection of utterances that necessitate manual annotation. Additionally, as time progresses, there may be a need to retrain these agents with new intents from the system-rejected queries to carry out additional tasks. Labeling all these emerging intents and rejected utterances over time is impractical, thus calling for an efficient mechanism to reduce annotation costs. In this paper, we introduce IDALC (Intent Detection and Active Learning based Correction), a semi-supervised framework designed to detect user intents and rectify system-rejected utterances while minimizing the need for human annotation. Empirical findings on various benchmark datasets demonstrate that our system surpasses baseline methods, achieving a 5-10% higher accuracy and a 4-8% improvement in macro-F1. Remarkably, we maintain the overall annotation cost at just 6-10% of the unlabelled data available to the system. The overall framework of IDALC is shown in Fig. 1

New intent discovery (NID) seeks to recognize both new and known intents from unlabeled user utterances, which finds prevalent use in practical dialogue systems. Existing works towards NID mainly adopt a cascaded architecture, wherein the first stage focuses on encoding the utterances into informative text embeddings beforehand, while the latter is to group similar embeddings into clusters (i.e., intents), typically by K-Means. However, such a cascaded pipeline fails to leverage the feedback from both steps for mutual refinement, and, meanwhile, the embedding-only clustering overlooks nuanced textual semantics, leading to suboptimal performance. To bridge this gap, this paper proposes NILC, a novel clustering framework specially catered for effective NID. Particularly, NILC follows an iterative workflow, in which clustering assignments are judiciously updated by carefully refining cluster centroids and text embeddings of uncertain utterances with the aid of large language models (LLMs). Specifically, NILC first taps into LLMs to create additional semantic centroids for clusters, thereby enriching the contextual semantics of the Euclidean centroids of embeddings. Moreover, LLMs are then harnessed to augment hard samples (ambiguous or terse utterances) identified from clusters via rewriting for subsequent cluster correction. Further, we inject supervision signals through non-trivial techniques seeding and soft must links for more accurate NID in the semi-supervised setting. Extensive experiments comparing NILC against multiple recent baselines under both unsupervised and semi-supervised settings showcase that NILC can achieve significant performance improvements over six benchmark datasets of diverse domains consistently.

Despite the effectiveness of quantization-aware training (QAT) in compressing deep neural networks, its performance on multi-task architectures often degrades significantly due to task-specific feature discrepancies and gradient conflicts. To address these challenges, we propose Gradient-Aware Balanced Feature Fusion (GABFusion), which dynamically balances gradient magnitudes and fuses task-specific features in a quantization-friendly manner. We further introduce Attention Distribution Alignment (ADA), a feature-level distillation strategy tailored for quantized models. Our method demonstrates strong generalization across network architectures and QAT algorithms, with theoretical guarantees on gradient bias reduction. Extensive experiments demonstrate that our strategy consistently enhances a variety of QAT methods across different network architectures and bit-widths. On PASCAL VOC and COCO datasets, the proposed approach achieves average mAP improvements of approximately 3.3% and 1.6%, respectively. When applied to YOLOv5 under 4-bit quantization, our method narrows the accuracy gap with the full-precision model to only 1.7% on VOC, showcasing its effectiveness in preserving performance under low-bit constraints. Notably, the proposed framework is modular, easy to integrate, and compatible with any existing QAT technique-enhancing the performance of quantized models without requiring modifications to the original network architecture.

Thinking Large Language Models (LLMs) generate explicit intermediate reasoning traces before final answers, potentially improving transparency, interpretability, and solution accuracy for code generation. However, the quality of these reasoning chains remains underexplored. We present a comprehensive empirical study examining the reasoning process and quality of thinking LLMs for code generation. We evaluate six state-of-the-art reasoning LLMs (DeepSeek-R1, OpenAI-o3-mini, Claude-3.7-Sonnet-Thinking, Gemini-2.0-Flash-Thinking, Gemini-2.5-Flash, and Qwen-QwQ) across 100 code generation tasks of varying difficulty from BigCodeBench. We quantify reasoning-chain structure through step counts and verbosity, conduct controlled step-budget adjustments, and perform a 21-participant human evaluation across three dimensions: efficiency, logical correctness, and completeness. Our step-count interventions reveal that targeted step increases can improve resolution rates for certain models/tasks, while modest reductions often preserve success on standard tasks, rarely on hard ones. Through systematic analysis, we develop a reasoning-problematic taxonomy, identifying completeness as the dominant failure mode. Task complexity significantly impacts reasoning quality; hard problems are substantially more prone to incompleteness than standard tasks. Our stability analysis demonstrates that thinking LLMs maintain consistent logical structures across computational effort levels and can self-correct previous errors. This study provides new insights into the strengths and limitations of current thinking LLMs in software engineering.