When Does AI Augment Work? A Workflow-Level Framework for Human-Agent Collaboration

By: AI Collaboration · Jiaying Wu · Caleb Ziems · Raymond Chan · Nancy F. Chen · Corlyss Chua · Gerard Chung · Jungpil Hahn · Wee Sun Lee · Zhengyuan Liu · Jamie Ng · Desmond C. Ong · Jeryl Ong · Da Ren Soon · Tianqi Song · Zhi-Xuan Tan · Sixing Tao · Emily Yang · Yajing Yang · Stella Xin Yin · Min-Yen Kan · Diyi Yang

14 September 2026 at 12:00

arXiv:2609.12482v1 Announce Type: new Abstract: We aim to characterise the value of artificial intelligence in the workplace. Current studies largely measure this value in terms of the current automation capabilities and public adoption of AI. However, such metrics ignore the greater impacts of human--agent collaboration in transforming the nature of work. To account for this, we must expand the scope of our analysis beyond atomised tasks of today, and instead focus on how AI can augment entire workflows of the future. To ground this analysis, we establish a precise definition of AI augmentation comprising six conditions, spanning durable net value, meaningful human control, accountability and recovery, and long-term human development through learning, career pathways, and job purpose. We elaborate on these conditions and apply the framework in a case study of AI-mediated social surveys. We conclude by outlining how organisations, researchers, and government leaders can use this framework to make sense of the future of work.

Circular RNA MCM3 recruits USP49 to stabilize PTBP1 and promote cisplatin resistance in cervical squamous cell carcinoma

Oncogene - Issue - nature.com science feeds

By: Chenchen Zhu · Ying Wei · Yuebo Li · Zhengzheng Chen · Ying Zhou · Hanyuan Liu

12 September 2026 at 08:00

Oncogene, Published online: 12 September 2026; doi:10.1038/s41388-026-03988-2

Circular RNA MCM3 recruits USP49 to stabilize PTBP1 and promote cisplatin resistance in cervical squamous cell carcinoma

FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models

cs.AI, q-bio.NC updates on arXiv.org

By: Yansen Han · Shengyi Liao · Peng Sun · Deyuan Liu · Yuanxing Zhang · Pengfei Wan · Tao Lin

10 September 2026 at 12:00

arXiv:2609.09905v1 Announce Type: cross Abstract: Preference alignment for flow and diffusion models now spans online reinforcement learning and offline preference optimization, but the relation between these methods remains unclear. In particular, existing forward-process alignment methods require fresh samples from the current model, while offline methods based on fixed preference pairs rely primarily on positive-only fine-tuning or DPO-style likelihood-ratio surrogates. We organize these approaches through a divergence-based framework and introduce FlowCPO, an offline forward-KL objective that uses both preferred and dispreferred samples without online rollouts. For linear interpolation, we show under explicit regularity conditions that the forward-KL objective is bounded by a contrastive flow matching loss, yielding a tractable surrogate on fixed data. We further show that this loss is nonnegative, whereas the signed regression loss of simplified FlowDPO can be unbounded below. In the in-domain setting, FlowCPO achieves higher mean GenEval and OCR scores than the evaluated baselines, reaching 0.84 and 0.87 versus 0.81 and 0.74 for FlowDPO at CFG 3.0. In the out-of-domain setting, the results are mixed, with the best GenEval result but lower reward scores than RFT on several metrics.

Foaming photopolymers as a high-resolution biomimetic printing platform

Nature - Issue - nature.com science feeds

By: Detao Qin · Xianghui Liu · Baian Kuang · Yuzhe Zhang · Masateru Ito · Tatsuya Katsuno · Yuan Liu · Jianduo Zhang · Claudia Rosa · Meifang Zhu · Mikihito Takenaka · Hiroshi Imahori · Ganesh N. Pandian · Taiki Yanagishima · Easan Sivaniah

9 September 2026 at 08:00

Nature, Published online: 09 September 2026; doi:10.1038/s41586-026-10968-9

Deep-foam photolithography uses light-controlled polymer foaming to create high-resolution, multifunctional microstructures with tunable optical, wetting and fluid-handling properties for advanced manufacturing applications.

Genomics and social practices at Mogou and other Gansu sites during prehistoric trans-Eurasian exchange

Cell

By: Han Shi · E. Andrew Bennett · Guoke Chen · Hui Wang · Tianyi Wang · Peng Cao · Ruilin Mao · Fan Bai · Qingyan Dai · Feng Liu · Xiaotian Feng · Yongan Wang · Chong Yao · Jing Wu · Wenjun Wang · Ming Zhang · Tianxiang Liu · Jingkun Ran · Yue-Chen Liu · Yuan Liu · Dan Ju · Chuyan Li · Hongwei Hou · Xinyi Liu · Qiaomei Fu

3 September 2026 at 08:00

Ancient DNA from 149 individuals at 11 sites in Gansu, China, dated to around 4,700–3,000 years ago, reveals human population history during early transcontinental exchanges of agriculture and technology, as well as contemporary social practices, at the large Mogou cemetery.

Test-Time Deep Thinking to Explore Implicit Rules

cs.AI, q-bio.NC updates on arXiv.org

By: Wentong Chen · Xin Cong · Zhong Zhang · Yaxi Lu · Siyuan Zhao · Yesai Wu · Qinyu Luo · Haotian Chen · Yankai Lin · Zhiyuan Liu · Maosong Sun

26 May 2026 at 12:00

arXiv:2605.24828v1 Announce Type: new Abstract: With the continuous advancement of Large Language Models (LLMs), intelligent agents are becoming increasingly vital. However, these agents often fail in environments governed by implicit rules--hidden constraints that cannot be observed directly and must be inferred through interaction. This causes agents to fall into repetitive trial-and-error loops, ultimately leading to task failure. To address this challenge, we propose Test-Time Exploration (TTExplore), a framework where a thinker component analyzes interaction history to infer these implicit rules and guide an actor. Effective exploration in this setting critically depends on the reasoning ability of the thinker. However, evaluating deep reasoning trajectories is inherently unstable and difficult, which poses a major obstacle to effective training. To overcome this issue, we introduce a novel and stable reinforcement learning pipeline. The core idea is to use accurate task-level scores as indirect rewards to bypass the difficulty of evaluating intermediate reasoning, and to retain only a single thinking node per trajectory to alleviate reward sparsity. Using this pipeline, we train a specialized 7B model, Exp-Thinker. Experiments on five text-based embodied tasks show that TTExplore equipped with Exp-Thinker improves baseline agent performance by an average of $14$-$19$ points, demonstrating the effectiveness of explicitly reasoning about implicit rules.

CITYREP: A Unified Benchmark for Urban Representations Across Cities, Tasks, and Modalities

cs.AI, q-bio.NC updates on arXiv.org

By: Junyuan Liu · Xinglei Wang · Zichao Zeng · Jiazhuang Feng · Quan Qin · Ilya Ilyankou · Guangsheng Dong · Tao Cheng

26 May 2026 at 12:00

arXiv:2605.26036v1 Announce Type: new Abstract: Urban representation learning encodes complex urban environments into general-purpose embeddings for diverse downstream tasks and emerging urban foundation models. However, current evaluations are limited, typically focusing on one or two cities and tasks and relying on random splits that introduce spatial leakage, leading to inflated performance and weak support for cross-location generalization and fair comparison. To address this, we propose CityRep, a unified benchmark that evaluates urban representations across data modalities, cities, and tasks using spatially structured splits. CityRep consists of three key components: (1) a spatial unit-agnostic evaluation framework that supports heterogeneous urban representations through a standardized alignment module; (2) a unified evaluation protocol using block-based spatial splits to mitigate spatial leakage and enable rigorous model comparison; and (3) an extensible multi-city, multi-task benchmark suite spanning 8 cities and 8 tasks across regression, classification, and distribution prediction. We evaluate 11 representative urban representation models. Results show that performance is highly sensitive to the split protocol, with random splits inflating scores and altering model rankings. We also observe substantial variability across cities and tasks, underscoring the need for generalization-aware evaluation. CityRep is released as a reproducible benchmark with datasets, evaluation pipelines, and diagnostic tools to facilitate fair comparison and support future research in urban representation learning towards urban foundation models.

MuNet: A Mutualistic Network for Joint 3D Human Mesh Recovery and 3D Clothed Human Reconstruction from Single Images

cs.AI, q-bio.NC updates on arXiv.org

By: Yunqi Gao · Leyuan Liu · Yuhan Li · Changxin Gao · Jingying Chen

26 May 2026 at 12:00

arXiv:2605.25861v2 Announce Type: cross Abstract: 3D human mesh recovery and 3D clothed human reconstruction are inherently related, yet they have long been studied in isolation, thereby overlooking the potential gains of joint optimization. To overcome this limitation, we propose to address these two tasks within a unified framework, which allows their mutual dependencies to be effectively exploited. Building on this idea, we propose MuNet, a mutualistic network for joint 3D human mesh recovery and 3D clothed human reconstruction from single images. First, we adopt 2-manifold graphs as a unified representation for all 3D models, enabling consistent modeling across 3D human mesh recovery and clothed human reconstruction. Second, we design an end-to-end graph convolutional network that progressively deforms an initial graph into a 3D human mesh and refines it into a detailed 3D clothed human model. Third, we introduce a mutualistic mechanism that allows reciprocal interaction between the two tasks {during training}, where 3D human mesh recovery provides guidance for 3D clothed human reconstruction, and reconstruction feedback refines the 3D human mesh recovery. We extensively evaluate MuNet on six benchmark datasets for 3D human mesh recovery and 3D clothed human reconstruction, including Human3.6M, 3DPW, MPI-INF-3DHP, THuman2.0, CAPE, and RenderPeople. Experimental results demonstrate that MuNet achieves state-of-the-art performance on both tasks across all datasets. The code of MuNet is released for research purposes at https://github.com/starVisionTeam/MuNet.

Data Difficulty and the Generalization--Extrapolation Tradeoff in LLM Fine-Tuning

cs.AI, q-bio.NC updates on arXiv.org

By: Siyuan Liu · Tinghong Chen · Xinghan Li · Yifei Wang · Jingzhao Zhang

26 May 2026 at 12:00

arXiv:2605.12906v2 Announce Type: replace-cross Abstract: Data selection during supervised fine-tuning (SFT) can critically change the behavior of large language models (LLMs). Although existing work has studied the effect of selecting data based on heuristics such as perplexity, difficulty, or length, the reported findings are often inconsistent or context-dependent. In this work, we systematically study the role of data difficulty in fine-tuning from both empirical and theoretical perspectives, and find that there is no universally optimal difficulty level; rather, its effectiveness depends on the dataset size. We show that for a fixed data budget, there exists an optimal data difficulty for SFT, and that this optimal difficulty shifts toward harder data as the data budget increases. To explain this phenomenon, we conduct controlled synthetic experiments that reveal a simple underlying mechanism: the interplay between the (in-distribution) generalization gap and the extrapolation gap. We further support this mechanism through a theoretical analysis using PAC-Bayesian generalization bounds. Overall, our results clarify how data size and difficulty jointly affect the trade-off between generalization and extrapolation in SFT, providing guidance for difficulty-based data selection under certain model and data conditions.

Graphicalized vision-language modeling for comprehensive lung nodule analysis and risk stratification

npj Digital Medicine

By: Danwen Zhao · Junfeng Xi · Xun Guo · Jintao Chai · Zhengshui Xu · Liang Li · Yan Xue · Qingyu Sun · Yinggang Zheng · Shiyuan Liu

11 April 2026 at 08:00

npj Digital Medicine, Published online: 11 April 2026; doi:10.1038/s41746-026-02602-9

Graphicalized vision-language modeling for comprehensive lung nodule analysis and risk stratification

Social Media Intervention Based on the Information-Motivation-Behavioral Skills Model Promotes HIV Testing and Reduces High-Risk Behaviors Among Men Who Have Sex With Men in Resource-Limited Settings in China: Randomized Controlled Trial

Journal of Medical Internet Research

By: Huanzhuo Mai · Jianyuan Liu · Liping Cheng · Hao Liang · Ping Cui · Qin Yao · Qin Cao · Heng Zhang · Wenping Liao · Shiran He · Xing Yang · Ping Cen · Jiegang Huang

8 April 2026 at 01:30

Background: Social media intervention may enhance HIV prevention among men who have sex with men, but the effect of this intervention in resource-limited settings remains unclear. Objective: This randomized controlled trial evaluated whether a social media intervention grounded in the information-motivation-behavioral skills (IMB) model could be beneficial for HIV prevention among men who have sex with men in resource-limited settings. Methods: Participants were recruited in Nanning, China, between April 2023 and April 2024. Eligible participants were randomly assigned to either the social media intervention group or the routine HIV prevention services control group. Participants in the intervention group received a 3-month social media intervention, which included completing video-based tasks. Baseline surveys were conducted, followed by follow-up surveys every 3 months, for a total of 2 follow-ups. Outcomes included HIV testing uptake, high-risk behavior, AIDS-related knowledge, safe sex self-efficacy, and attitude. Results: A total of 180 eligible men who have sex with men were enrolled (90 per group). Follow-up rates were 97.8% (88/90) and 95.5% (86/90) for the intervention and control groups, respectively. At the follow-ups, the intervention group demonstrated significantly higher uptake of HIV testing, a lower proportion of participants reporting high-risk sexual behaviors, and higher condom use self-efficacy compared to the control group (all

Generation Is Compression: Zero-Shot Video Coding via Stochastic Rectified Flow

cs.AI, q-bio.NC updates on arXiv.org

By: Ziyue Zeng · Xun Su · Haoyuan Liu · Bingyu Lu · Yui Tatsumi · Hiroshi Watanabe

3 April 2026 at 12:00

arXiv:2603.26571v2 Announce Type: replace-cross Abstract: Recent advances in generative modeling have enabled perceptual video compression at ultra-low bitrates, yet existing methods predominantly treat the generative model as a refinement or reconstruction module attached to a separately designed codec backbone. We propose \emph{Generative Video Codebook Codec} (GVCC), a zero-shot framework that turns a pretrained video generative model into the codec itself: the transmitted bitstream directly specifies the generative decoding trajectory, with no retraining required. To enable this, we convert the deterministic rectified-flow ODE of modern video foundation models into an equivalent SDE at inference time, unlocking per-step stochastic injection points for codebook-driven compression. Building on this unified backbone, we instantiate three complementary conditioning strategies -- \emph{Image-to-Video} (I2V) with autoregressive GOP chaining, tail latent residual correction, and adaptive atom allocation, \emph{Text-to-Video} (T2V) operating at near-zero side information as a pure generative prior, and \emph{First-Last-Frame-to-Video} (FLF2V) with boundary-sharing GOP chaining for dual-anchor temporal control. Together, these variants span a principled trade-off space between spatial fidelity, temporal coherence, and compression efficiency. Experiments on standard benchmarks show that GVCC achieves high-quality reconstruction below 0.002\,bpp while supporting flexible bitrate control through a single hyperparameter.

Genetically encoded fluorescent reporters to visualize α-synuclein pathology in live brain

Cell

By: Li Zhang · Minhui Yu · Guoqing Chen · Siyuan Ge · Mengdi Wang · Xianying Zhang · Miao Zhao · Huating Gu · Meizhu Huang · Aixue Liu · Gengxin Ran · Zeyuan Liu · Tiepeng Liao · Qi Chen · Chenjian Miao · Yao Lu · Yibing Wang · Fengchao Wang · Zhihui Liu · Hongying Zhu · Qian Wu · Zhuan Zhou · Wei Xiong · Xiaoqun Wang · Peng Cao

4 March 2026 at 08:00

The development of genetically encoded fluorescent reporters, along with their corresponding knock-in mouse lines for labeling α-Syn inclusions, enables diverse applications in studying the propagation and pathological effects of α-Syn inclusions in the live brain.

PSPA-Bench: A Personalized Benchmark for Smartphone GUI Agent

cs.AI, q-bio.NC updates on arXiv.org

By: Hongyi Nie · Xunyuan Liu · Yudong Bai · Yaqing Wang · Yang Liu · Quanming Yao · Zhen Wang

1 April 2026 at 12:00

arXiv:2603.29318v1 Announce Type: new Abstract: Smartphone GUI agents execute tasks by operating directly on app interfaces, offering a path to broad capability without deep system integration. However, real-world smartphone use is highly personalized: users adopt diverse workflows and preferences, challenging agents to deliver customized assistance rather than generic solutions. Existing GUI agent benchmarks cannot adequately capture this personalization dimension due to sparse user-specific data and the lack of fine-grained evaluation metrics. To address this gap, we present PSPA-Bench, the benchmark dedicated to evaluating personalization in smartphone GUI agents. PSPA-Bench comprises over 12,855 personalized instructions aligned with real-world user behaviors across 10 representative daily-use scenarios and 22 mobile apps, and introduces a structure-aware process evaluation method that measures agents' personalized capabilities at a fine-grained level. Through PSPA-Bench, we benchmark 11 state-of-the-art GUI agents. Results reveal that current methods perform poorly under personalized settings, with even the strongest agent achieving limited success. Our analysis further highlights three directions for advancing personalized GUI agents: (1) reasoning-oriented models consistently outperform general LLMs, (2) perception remains a simple yet critical capability, and (3) reflection and long-term memory mechanisms are key to improving adaptation. Together, these findings establish PSPA-Bench as a foundation for systematic study and future progress in personalized GUI agents.

Accelerating Diffusion Large Language Models with SlowFast Sampling: The Three Golden Principles

cs.AI, q-bio.NC updates on arXiv.org

By: Qingyan Wei · Yaojie Zhang · Zhiyuan Liu · Puyu Zeng · Yuxuan Wang · Biqing Qi · Dongrui Liu · Linfeng Zhang

1 April 2026 at 12:00

arXiv:2506.10848v3 Announce Type: replace-cross Abstract: Diffusion-based language models (dLLMs) have emerged as a promising alternative to traditional autoregressive LLMs by enabling parallel token generation and significantly reducing inference latency. However, existing sampling strategies for dLLMs, such as confidence-based or semi-autoregressive decoding, often suffer from static behavior, leading to suboptimal efficiency and limited flexibility. In this paper, we propose SlowFast Sampling, a novel dynamic sampling strategy that adaptively alternates between exploratory and accelerated decoding stages. Our method is guided by three golden principles: certainty principle, convergence principle, and positional principle, which govern when and where tokens can be confidently and efficiently decoded. We further integrate our strategy with dLLM-Cache to reduce redundant computation. Extensive experiments across benchmarks and models show that SlowFast Sampling achieves up to 15.63$\times$ speedup on LLaDA with minimal accuracy drop, and up to 34.22$\times$ when combined with caching. Notably, our approach outperforms strong autoregressive baselines like LLaMA3 8B in throughput, demonstrating that well-designed sampling can unlock the full potential of dLLMs for fast and high-quality generation.

Gene regulatory landscape dissected by single-cell four-omics sequencing

Nature - Issue - nature.com science feeds

By: Yujie Chen · Zhiyuan Liu · Heming Xu · Jiayu Liu · Mengxuan Wang · Yi Chi · Boyuan Liang · Menghan Liu · Yongli Peng · Hao Ge · Dong Xing

1 April 2026 at 08:00

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10322-z

Combining single-cell parallel profiling of genome conformation, histone modifications, chromatin accessibility and gene expression reveals dynamics and intranuclear spatial clustering of epigenome profiles, enabling sophisticated analysis of the regulatory landscape across cell types and tissues.

Cellular Senescence in Gastric Cancer: Molecular Mechanisms, Microenvironment Remodeling and Therapeutic Implications

Omics in Gastric

By: Zhiyuan Shi · Zhao Sun · Yuan Liu · Yuping Ge · Ning Jia · Yuwei Hua · Lin Zhao · Xiang Wang · Yi Ba

30 March 2026 at 18:00

Aging Dis. 2026 Mar 19. doi: 10.14336/AD.2025.1571. Online ahead of print.

ABSTRACT

Gastric cancer (GC) remains a leading cause of cancer-related morbidity and mortality worldwide, with poor prognosis for advanced-stage patients. Therefore, in-depth exploration of the mechanisms underlying GC initiation and progression, as well as the development of novel therapeutic strategies, is of crucial importance. Cellular senescence is a stable cell cycle arrest program that plays a dual role in GC. It exerts tumor-suppressive effects via growth arrest but also promotes tumor progression and immune evasion by remodeling the tumor microenvironment (TME) through senescence-associated secretory phenotype (SASP). This review comprehensively elucidates the molecular mechanisms of cellular senescence in GC and the core regulatory networks involving gene regulation, epigenetic modifications, metabolic reprogramming, and cell cycle arrest. Additionally, the review highlights how senescent cells foster an immunosuppressive microenvironment via SASP, forming a self-reinforcing feed-forward loop. Regarding therapeutic strategies, we summarize potential approaches targeting cellular senescence, including senescence induction, senescent cell clearance, SASP modulation, and multi-target synergistic therapy by integrating epigenetic regulation, metabolic intervention, and immune microenvironment modulation. Despite progress, numerous challenges remain. Future studies should leverage multi-omics technologies, novel models' development, and large-scale clinical trials to advance the clinical translation of GC cellular senescence research, providing new insights for improving prognosis.

PMID:41910653 | DOI:10.14336/AD.2025.1571

Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation

cs.AI, q-bio.NC updates on arXiv.org

By: Yichen Zhang · Da Peng · Zonghao Guo · Zijian Zhang · Xuesong Yang · Tong Sun · Shichu Sun · Yidan Zhang · Yanghao Li · Haiyan Zhao · Wang Xu · Qi Shi · Yangang Sun · Chi Chen · Shuo Wang · Yukun Yan · Xu Han · Qiang Ma · Wei Ke · Liang Wang · Zhiyuan Liu · Maosong Sun

16 March 2026 at 12:00

arXiv:2603.12793v1 Announce Type: cross Abstract: A recent cutting-edge topic in multimodal modeling is to unify visual comprehension and generation within a single model. However, the two tasks demand mismatched decoding regimes and visual representations, making it non-trivial to jointly optimize within a shared feature space. In this work, we present Cheers, a unified multimodal model that decouples patch-level details from semantic representations, thereby stabilizing semantics for multimodal understanding and improving fidelity for image generation via gated detail residuals. Cheers includes three key components: (i) a unified vision tokenizer that encodes and compresses image latent states into semantic tokens for efficient LLM conditioning, (ii) an LLM-based Transformer that unifies autoregressive decoding for text generation and diffusion decoding for image generation, and (iii) a cascaded flow matching head that decodes visual semantics first and then injects semantically gated detail residuals from the vision tokenizer to refine high-frequency content. Experiments on popular benchmarks demonstrate that Cheers matches or surpasses advanced UMMs in both visual understanding and generation. Cheers also achieves 4x token compression, enabling more efficient high-resolution image encoding and generation. Notably, Cheers outperforms the Tar-1.5B on the popular benchmarks GenEval and MMBench, while requiring only 20% of the training cost, indicating effective and efficient (i.e., 4x token compression) unified multimodal modeling. We will release all code and data for future research.

LatentChem: From Textual CoT to Latent Thinking in Chemical Reasoning

cs.AI, q-bio.NC updates on arXiv.org

By: Xinwu Ye · Yicheng Mao · Jia Zhang · Yimeng Liu · Li Hao · Fang Wu · Zhiwei Li · Yuxuan Liao · Zehong Wang · Yingcheng Wu · Zhiyuan Liu · Zhenfei Yin · Li Yuan · Philip Torr · Huan Sun · Xiangxiang Zeng · Mengdi Wang · Le Cong · Shenghua Gao · Xiangru Tang

16 March 2026 at 12:00

arXiv:2602.07075v4 Announce Type: replace-cross Abstract: Chemical large language models (LLMs) predominantly rely on explicit Chain-of-Thought (CoT) in natural language to perform complex reasoning. However, chemical reasoning is inherently continuous and structural, and forcing it into discrete linguistic tokens introduces a fundamental representation mismatch that constrains both efficiency and performance. We introduce LatentChem, a latent reasoning interface that decouples chemical computation from textual generation, enabling models to perform multi-step reasoning directly in continuous latent space while emitting language only for final outputs. Remarkably, we observe a consistent emergent behavior: when optimized solely for task success, models spontaneously internalize reasoning, progressively abandoning verbose textual derivations in favor of implicit latent computation. This shift is not merely stylistic but computationally advantageous. Across diverse chemical reasoning benchmarks, LatentChem achieves a 59.88\% non-tie win rate over strong CoT-based baselines on ChemCoTBench, while delivering a 10.84$\times$ average reduction in reasoning overhead. Our results provide empirical evidence that chemical reasoning is more naturally and effectively realized as continuous latent dynamics rather than discretized linguistic trajectories.

Evo: Autoregressive-Diffusion Large Language Models with Evolving Balance

cs.AI, q-bio.NC updates on arXiv.org

By: Junde Wu · Minhao Hu · Jiayuan Zhu · Yuyuan Liu · Tianyi Zhang · Kang Li · Jingkun Chen · Jiazhen Pan · Min Xu · Yueming Jin

10 March 2026 at 12:00

arXiv:2603.06617v1 Announce Type: cross Abstract: We introduce \textbf{Evo}, a duality latent trajectory model that bridges autoregressive (AR) and diffusion-based language generation within a continuous evolutionary generative framework. Rather than treating AR decoding and diffusion generation as separate paradigms, Evo reconceptualizes text generation as a latent flow: each token is associated with a vector-valued embedding that evolves over a progression variable $t_i \in [0, 1]$, indicating its semantic maturity. Low $t_i$ values correspond to confident AR-like refinement, while high values invoke diffusion-style planning, allowing the model to adaptively balance AR and diffusion based on uncertainty. Theoretically, we show that both AR and diffusion models emerge as discretizations of a shared probability flow, and we derive Evo's training objective from a unified variational ELBO. The model is implemented as a time-conditioned Transformer governed by a shared vector field, trained end-to-end to jointly infer latent codes and their progression times. During decoding, Evo performs efficient, semantics-aware refinement, achieving high-quality outputs without sacrificing speed. Empirically, Evo 8B achieves state-of-the-art or highly competitive results on 15 diverse benchmarks, including reasoning (GSM8K, ARC-C), code generation (HumanEval, MBPP), and general language understanding, while maintaining fast inference speed. Our results demonstrate that Evo delivers a new paradigm for LLM design with strong generation quality, robust symbolic reasoning, and decoding efficiency.

Reading view