❌

Normal view

Received — 6 April 2026 ⏭

Unmasking FCGR2B as a high-grade serous ovarian cancer specific marker of immune suppression and tumor progression through multi-omics mining

✇(Multiomics OR Omics) AND (Pancreatic)
By: Jialu Zhou · Tao Zeng · Mingxia Ye · Mingxia Li · Liang Wen · Xinyi Liu · Zhe Zhang · Yuanguang Meng
4 April 2026 at 18:00

Transl Oncol. 2026 Apr 3;67:102748. doi: 10.1016/j.tranon.2026.102748. Online ahead of print.

ABSTRACT

BACKGROUND: Epithelial ovarian cancer (EOC) encompasses five major histological subtypes with marked genetic, immunological, and clinical heterogeneity. While genome-wide association studies (GWAS) have identified subtype-specific risk loci, a critical gap remains in understanding how plasma proteins influence immune-cell traits and contribute to EOC pathogenesis.

METHODS: We integrated subtype-stratified GWAS data from two EOC cohorts with plasma proteomics and immune-cell traits to construct protein-immune-EOC regulatory landscapes using a three-stage Mendelian randomization framework. Single-cell RNA-seq and multiplex immunofluorescence were employed to delineate the cellular distribution and spatial context of causal proteins. Subsequent analyses characterized immune infiltration, macrophage polarization, and clinicopathological associations. Drug-gene correlations were used to identify potential therapeutic targets, and transcriptomic analyses were applied to delineate the underlying transcriptional landscape.

RESULTS: We identified 20 subtype-specific protein-immune-EOC regulatory axes, with FCGR2B emerging as a causal plasma protein in immune regulation and high-grade serous ovarian cancer (HGSOC) progression. FCGR2B was highly expressed in tumor-associated macrophages and was associated with an M2-like polarization phenotype. Functional characterization revealed that FCGR2B was associated with shorter progression-free survival and an immunosuppressive tumor microenvironment. Transcriptomic analyses revealed altered NF-κB signaling upon FCGR2B knockdown, and drug-response data suggested a potential association between high FCGR2B expression and sensitivity to NF-κB inhibitors.

CONCLUSIONS: These findings delineate subtype-specific genetically informed protein-immune regulatory landscapes in EOC and identify FCGR2B as a key immunoregulatory and prognostic biomarker in HGSOC, suggesting FCGR2B as a potential therapeutic vulnerability that warrants further investigation.

PMID:41934917 | DOI:10.1016/j.tranon.2026.102748

Received — 4 April 2026 ⏭

Infeasibility Aware Large Language Models for Combinatorial Optimization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yakun Wang · Min Chen · Zeguan Wu · Junyu Liu · Sitao Zhang · Zhenwen Shao
3 April 2026 at 12:00
arXiv:2604.01455v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly explored for NP-hard combinatorial optimization problems, but most existing methods emphasize feasible-instance solution generation and do not explicitly address infeasibility detection. We propose an infeasibility-aware framework that combines certifiable dataset construction, supervised fine-tuning, and LLM-assisted downstream search. For the minor-embedding problem, we introduce a new mathematical programming formulation together with provable zero-phase infeasibility screening, which enables scalable construction of training instances labeled either as feasible with structured certificates or as certifiably infeasible. Using training data generated through this exact optimization pipeline, we show that an 8B-parameter LLM can be fine-tuned to jointly perform solution generation and infeasibility detection. We further utilize LLM outputs as warm starts for downstream local search, providing a practical way to accelerate optimization even when the LLM outputs are imperfect. Experiments show that our fine-tuned model improves overall accuracy by up to 30\% over GPT-5.2; meanwhile LLM-guided warm starts provide up to $2\times$ speedup compared with starting from scratch in downstream local search.

Exploring Robust Multi-Agent Workflows for Environmental Data Management

✇cs.AI, q-bio.NC updates on arXiv.org
By: Boyuan Guan · Jason Liu · Yanzhao Wu · Kiavash Bahreini
3 April 2026 at 12:00
arXiv:2604.01647v1 Announce Type: new Abstract: Embedding LLM-driven agents into environmental FAIR data management is compelling - they can externalize operational knowledge and scale curation across heterogeneous data and evolving conventions. However, replacing deterministic components with probabilistic workflows changes the failure mode: LLM pipelines may generate plausible but incorrect outputs that pass superficial checks and propagate into irreversible actions such as DOI minting and public release. We introduce EnviSmart, a production data management system deployed on campus-wide storage infrastructure for environmental research. EnviSmart treats reliability as an architectural property through two mechanisms: a three-track knowledge architecture that externalizes behaviors (governance constraints), domain knowledge (retrievable context), and skills (tool-using procedures) as persistent, interlocking artifacts; and a role-separated multi-agent design where deterministic validators and audited handoffs restore fail-stop semantics at trust boundaries before irreversible steps. We compare two production deployments. The University's GIS Center Ecological Archive (849 curated datasets) serves as a single-agent baseline. SF2Bench, a compound flooding benchmark comprising 2,452 monitoring stations and 8,557 published files spanning 39 years, validates the multi-agent workflow. The multi-agent approach improved both efficiency - completed by a single operator in two days with repeated artifact reuse across deployments - and reliability: audited handoffs detected and blocked a coordinate transformation error affecting all 2,452 stations before publication. A representative incident (ISS-004) demonstrated boundary-based containment with 10-minute detection latency, zero user exposure, and 80-minute resolution. This paper has been accepted at PEARC 2026.

Robust Embodied Perception in Dynamic Environments via Disentangled Weight Fusion

✇cs.AI, q-bio.NC updates on arXiv.org
By: Juncen Guo · Xiaoguang Zhu · Jingyi Wu · Jingyu Zhang · Jingnan Cai · Zhenghao Niu · Liang Song
3 April 2026 at 12:00
arXiv:2604.01669v1 Announce Type: cross Abstract: Embodied perception systems face severe challenges of dynamic environment distribution drift when they continuously interact in open physical spaces. However, the existing domain incremental awareness methods often rely on the domain id obtained in advance during the testing phase, which limits their practicability in unknown interaction scenarios. At the same time, the model often overfits to the context-specific perceptual noise, which leads to insufficient generalization ability and catastrophic forgetting. To address these limitations, we propose a domain-id and exemplar-free incremental learning framework for embodied multimedia systems, which aims to achieve robust continuous environment adaptation. This method designs a disentangled representation mechanism to remove non-essential environmental style interference, and guide the model to focus on extracting semantic intrinsic features shared across scenes, thereby eliminating perceptual uncertainty and improving generalization. We further use the weight fusion strategy to dynamically integrate the old and new environment knowledge in the parameter space, so as to ensure that the model adapts to the new distribution without storing historical data and maximally retains the discrimination ability of the old environment. Extensive experiments on multiple standard benchmark datasets show that the proposed method significantly reduces catastrophic forgetting in a completely exemplar-free and domain-id free setting, and its accuracy is better than the existing state-of-the-art methods.

Bridging Large-Model Reasoning and Real-Time Control via Agentic Fast-Slow Planning

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jiayi Chen · Shuai Wang · Guangxu Zhu · Chengzhong Xu
3 April 2026 at 12:00
arXiv:2604.01681v1 Announce Type: cross Abstract: Large foundation models enable powerful reasoning for autonomous systems, but mapping semantic intent to reliable real-time control remains challenging. Existing approaches either (i) let Large Language Models (LLMs) generate trajectories directly - brittle, hard to verify, and latency-prone - or (ii) adjust Model Predictive Control (MPC) objectives online - mixing slow deliberation with fast control and blurring interfaces. We propose Agentic Fast-Slow Planning, a hierarchical framework that decouples perception, reasoning, planning, and control across natural timescales. The framework contains two bridges. Perception2Decision compresses scenes into ego-centric topologies using an on-vehicle Vision-Language Model (VLM) detector, then maps them to symbolic driving directives in the cloud with an LLM decision maker - reducing bandwidth and delay while preserving interpretability. Decision2Trajectory converts directives into executable paths: Semantic-Guided A* embeds language-derived soft costs into classical search to bias solutions toward feasible trajectories, while an Agentic Refinement Module adapts planner hyperparameters using feedback and memory. Finally, MPC tracks the trajectories in real time, with optional cloud-guided references for difficult cases. Experiments in CARLA show that Agentic Fast-Slow Planning improves robustness under perturbations, reducing lateral deviation by up to 45% and completion time by over 12% compared to pure MPC and an A*-guided MPC baseline. Code is available at https://github.com/cjychenjiayi/icra2026_AFSP.

DriveDreamer-Policy: A Geometry-Grounded World-Action Model for Unified Generation and Planning

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yang Zhou · Xiaofeng Wang · Hao Shao · Letian Wang · Guosheng Zhao · Jiangnan Shao · Jiagang Zhu · Tingdong Yu · Zheng Zhu · Guan Huang · Steven L. Waslander
3 April 2026 at 12:00
arXiv:2604.01765v1 Announce Type: cross Abstract: Recently, world-action models (WAM) have emerged to bridge vision-language-action (VLA) models and world models, unifying their reasoning and instruction-following capabilities and spatio-temporal world modeling. However, existing WAM approaches often focus on modeling 2D appearance or latent representations, with limited geometric grounding-an essential element for embodied systems operating in the physical world. We present DriveDreamer-Policy, a unified driving world-action model that integrates depth generation, future video generation, and motion planning within a single modular architecture. The model employs a large language model to process language instructions, multi-view images, and actions, followed by three lightweight generators that produce depth, future video, and actions. By learning a geometry-aware world representation and using it to guide both future prediction and planning within a unified framework, the proposed model produces more coherent imagined futures and more informed driving actions, while maintaining modularity and controllable latency. Experiments on the Navsim v1 and v2 benchmarks demonstrate that DriveDreamer-Policy achieves strong performance on both closed-loop planning and world generation tasks. In particular, our model reaches 89.2 PDMS on Navsim v1 and 88.7 EPDMS on Navsim v2, outperforming existing world-model-based approaches while producing higher-quality future video and depth predictions. Ablation studies further show that explicit depth learning provides complementary benefits to video imagination and improves planning robustness.

Omni123: Exploring 3D Native Foundation Models with Limited 3D Data by Unifying Text to 2D and 3D Generation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Chongjie Ye · Cheng Cao · Chuanyu Pan · Yiming Hao · Yihao Zhi · Yuanming Hu · Xiaoguang Han
3 April 2026 at 12:00
arXiv:2604.02289v1 Announce Type: cross Abstract: Recent multimodal large language models have achieved strong performance in unified text and image understanding and generation, yet extending such native capability to 3D remains challenging due to limited data. Compared to abundant 2D imagery, high-quality 3D assets are scarce, making 3D synthesis under-constrained. Existing methods often rely on indirect pipelines that edit in 2D and lift results into 3D via optimization, sacrificing geometric consistency. We present Omni123, a 3D-native foundation model that unifies text-to-2D and text-to-3D generation within a single autoregressive framework. Our key insight is that cross-modal consistency between images and 3D can serve as an implicit structural constraint. By representing text, images, and 3D as discrete tokens in a shared sequence space, the model leverages abundant 2D data as a geometric prior to improve 3D representations. We introduce an interleaved X-to-X training paradigm that coordinates diverse cross-modal tasks over heterogeneous paired datasets without requiring fully aligned text-image-3D triplets. By traversing semantic-visual-geometric cycles (e.g., text to image to 3D to image) within autoregressive sequences, the model jointly enforces semantic alignment, appearance fidelity, and multi-view geometric consistency. Experiments show that Omni123 significantly improves text-guided 3D generation and editing, demonstrating a scalable path toward multimodal 3D world models.

DR-LoRA: Dynamic Rank LoRA for Fine-Tuning Mixture-of-Experts Models

✇cs.AI, q-bio.NC updates on arXiv.org
By: Guanzhi Deng · Bo Li · Ronghao Chen · Xiujin Liu · Zhuo Han · Huacan Wang · Lijie Wen · Linqi Song
3 April 2026 at 12:00
arXiv:2601.04823v5 Announce Type: replace Abstract: Mixture-of-Experts (MoE) has become a prominent paradigm for scaling Large Language Models (LLMs). Parameter-efficient fine-tuning methods, such as LoRA, are widely adopted to adapt pretrained MoE LLMs to downstream tasks. However, existing approaches typically assign identical LoRA ranks to all expert modules, ignoring the heterogeneous specialization of pretrained experts. This uniform allocation leads to a resource mismatch: task-relevant experts are under-provisioned, while less relevant ones receive redundant parameters. To address this, we propose DR-LoRA, a Dynamic Rank LoRA framework for fine-tuning pretrained MoE models. Specifically, DR-LoRA initializes all expert LoRA modules with a small active rank and uses an expert saliency score, which combines routing frequency and gradient-based rank importance, to identify which experts would benefit most from additional capacity. It then periodically expands the active ranks of the task-critical expert LoRA, progressively constructing a heterogeneous rank distribution tailored to the target task. Experiments on three MoE models across six tasks show that DR-LoRA consistently outperforms LoRA and other strong baselines, demonstrating that task-adaptive heterogeneous rank allocation is an effective strategy to improve active capacity utilization in MoE fine-tuning.

Unified Optimization of Source Weights and Transfer Quantities in Multi-Source Transfer Learning: An Asymptotic Framework

✇cs.AI, q-bio.NC updates on arXiv.org
By: Qingyue Zhang · Chang Chu · Haohao Fu · Tianren Peng · Yanru Wu · Guanbo Huang · Yang Li · Shao-Lun Huang
3 April 2026 at 12:00
arXiv:2601.10779v2 Announce Type: replace-cross Abstract: In multi-source transfer learning, a key challenge lies in how to appropriately differentiate and utilize heterogeneous source tasks. However, existing multi-source methods typically focus on optimizing either the source weights or the amount of transferred samples, largely neglecting their joint consideration. In this work, we propose a theoretical framework, Unified Optimization of Weights and Quantities (UOWQ), that jointly determines the optimal source weights and transfer quantities for each source task. Specifically, the framework formulates multi-source transfer learning as a parameter estimation problem based on an asymptotic analysis of a Kullback--Leibler divergence--based generalization error measure, leading to two main theoretical findings: 1) using all available source samples is always optimal when the weights are properly adjusted; 2) the optimal source weights are characterized by a principled optimization problem whose structure explicitly incorporates the Fisher information, parameter discrepancy, parameter dimensionality, and transfer quantities. Building on the theoretical results, we further propose a practical algorithm for multi-source transfer learning, and extend it to multi-task learning settings where each task simultaneously serves as both a source and a target. Extensive experiments on real-world benchmarks, including DomainNet and Office-Home, demonstrate that UOWQ consistently outperforms strong baselines. The results validate both the theoretical predictions and the practical effectiveness of our framework.

MOON3.0: Reasoning-aware Multimodal Representation Learning for E-commerce Product Understanding

✇cs.AI, q-bio.NC updates on arXiv.org
By: Junxian Wu · Chenghan Fu · Zhanheng Nie · Daoze Zhang · Bowen Wan · Wanxian Guan · Chuan Yu · Jian Xu · Bo Zheng
3 April 2026 at 12:00
arXiv:2604.00513v2 Announce Type: replace-cross Abstract: With the rapid growth of e-commerce, exploring general representations rather than task-specific ones has attracted increasing attention. Although recent multimodal large language models (MLLMs) have driven significant progress in product understanding, they are typically employed as feature extractors that implicitly encode product information into global embeddings, thereby limiting their ability to capture fine-grained attributes. Therefore, we argue that leveraging the reasoning capabilities of MLLMs to explicitly model fine-grained product attributes holds significant potential. Nevertheless, achieving this goal remains non-trivial due to several key challenges: (i) long-context reasoning tends to dilute the model's attention to salient information in the raw input; (ii) supervised fine-tuning (SFT) primarily encourages rigid imitation, limiting the exploration of effective reasoning strategies; and (iii) fine-grained details are progressively attenuated during forward propagation. To address these issues, we propose MOON3.0, the first reasoning-aware MLLM-based model for product representation learning. Our method (1) employs a multi-head modality fusion module to adaptively integrate raw signals; (2) incorporates a joint contrastive and reinforcement learning framework to autonomously explore more effective reasoning strategies; and (3) introduces a fine-grained residual enhancement module to progressively preserve local details throughout the network. Additionally, we release a large-scale multimodal e-commerce benchmark MBE3.0. Experimentally, our model demonstrates state-of-the-art zero-shot performance across various downstream tasks on both our benchmark and public datasets.

UniMixer: A Unified Architecture for Scaling Laws in Recommendation Systems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Mingming Ha · Guanchen Wang · Linxun Chen · Xuan Rao · Yuexin Shi · Tianbao Ma · Zhaojie Liu · Yunqian Fan · Zilong Lu · Yanan Niu · Han Li · Kun Gai
3 April 2026 at 12:00
arXiv:2604.00590v2 Announce Type: replace-cross Abstract: In recent years, the scaling laws of recommendation models have attracted increasing attention, which govern the relationship between performance and parameters/FLOPs of recommenders. Currently, there are three mainstream architectures for achieving scaling in recommendation models, namely attention-based, TokenMixer-based, and factorization-machine-based methods, which exhibit fundamental differences in both design philosophy and architectural structure. In this paper, we propose a unified scaling architecture for recommendation systems, namely \textbf{UniMixer}, to improve scaling efficiency and establish a unified theoretical framework that unifies the mainstream scaling blocks. By transforming the rule-based TokenMixer to an equivalent parameterized structure, we construct a generalized parameterized feature mixing module that allows the token mixing patterns to be optimized and learned during model training. Meanwhile, the generalized parameterized token mixing removes the constraint in TokenMixer that requires the number of heads to be equal to the number of tokens. Furthermore, we establish a unified scaling module design framework for recommender systems, which bridges the connections among attention-based, TokenMixer-based, and factorization-machine-based methods. To further boost scaling ROI, a lightweight UniMixing module is designed, \textbf{UniMixing-Lite}, which further compresses the model parameters and computational cost while significantly improve the model performance. The scaling curves are shown in the following figure. Extensive offline and online experiments are conducted to verify the superior scaling abilities of \textbf{UniMixer}.

ORBIT: Scalable and Verifiable Data Generation for Search Agents on a Tight Budget

✇cs.AI, q-bio.NC updates on arXiv.org
By: Nandan Thakur · Zijian Chen · Xueguang Ma · Jimmy Lin
3 April 2026 at 12:00
arXiv:2604.01195v2 Announce Type: replace-cross Abstract: Search agents, which integrate language models (LMs) with web search, are becoming crucial for answering complex user queries. Constructing training datasets for deep research tasks, involving multi-step retrieval and reasoning, remains challenging due to expensive human annotation, or cumbersome prerequisites. In this work, we introduce ORBIT, a training dataset with 20K reasoning-intensive queries with short verifiable answers, generated using a frugal framework without relying on paid API services. The modular framework relies on four stages: seed creation, question-answer pair generation, and two stages of verification: self and external. ORBIT spans 15 domains and each training pair requires 4-5 reasoning steps, with external search verification required from the complete web. We train Qwen3-4B as the base model on ORBIT using GRPO and evaluate it on Wikipedia question answering tasks. Extensive experiment results demonstrate that ORBIT-4B achieves strong performance among sub-4B LLMs as search agents, proving the utility of synthetic datasets. Our framework, code and datasets are open-sourced and available publicly.

User-preference alignment with uncertainty-aware interactive rectification for liver organ and tumor segmentation and analysis from CT images

✇npj Digital Medicine
By: Guangyuan Zhao · Yang Wang · Chen Gong · Zipei Wang · Guobin Huang · Xuechun Zhao · Tengfei Chao · Bo Yang
3 April 2026 at 08:00

npj Digital Medicine, Published online: 03 April 2026; doi:10.1038/s41746-026-02544-2

User-preference alignment with uncertainty-aware interactive rectification for liver organ and tumor segmentation and analysis from CT images

Ferritin aggregation cell engager for CAR T avidity engineering against refractory leukemias

✇Cell
By: Feng Li · Yuxing Hu · Yan Wang · Xiao Zhang · Shaowei Qiu · Xiaoming Hu · Wenjing Wang · Shuang Wang · Peilin Guo · Cui Song · Huaiji Wang · Changhao He · Jie Sun · Di Yu · Shuhua Yi · Jun Wei · Feifan Liu · Caopei Zheng · Jianxiang Wang · Yuhua Li · Guanghui Ma · Wei Wei
9 March 2026 at 08:00
Li et al. developed a ferritin aggregation cell engager that helps CAR T cells better recognize and attack leukemia cells without re-engineering the CAR itself. This versatile platform overcomes antigen modulation and enables combination with chemotherapy.

Unified modeling of 3D molecular generation via atomic interactions with PocketXMol

✇Cell
By: Xingang Peng · Ruihan Guo · Fenglin Guo · Ziyi Wang · Jiayu Sun · Jiaqi Guan · Yinjun Jia · Yan Xu · Yanwen Huang · Muhan Zhang · Jian Peng · Xinquan Wang · Chuanhui Han · Zihua Wang · Jianzhu Ma
18 February 2026 at 08:00
A versatile, atom-level generative AI model enables unified pocket-interacting tasks, from docking to de novo design, and demonstrates robust experimental validation for both small-molecule and peptide therapeutics.
Received — 2 April 2026 ⏭

PAR$^2$-RAG: Planned Active Retrieval and Reasoning for Multi-Hop Question Answering

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xingyu Li · Rongguang Wang · Yuying Wang · Mengqing Guo · Chenyang Li · Tao Sheng · Sujith Ravi · Dan Roth
1 April 2026 at 12:00
arXiv:2603.29085v1 Announce Type: new Abstract: Large language models (LLMs) remain brittle on multi-hop question answering (MHQA), where answering requires combining evidence across documents through retrieval and reasoning. Iterative retrieval systems can fail by locking onto an early low-recall trajectory and amplifying downstream errors, while planning-only approaches may produce static query sets that cannot adapt when intermediate evidence changes. We propose \textbf{Planned Active Retrieval and Reasoning RAG (PAR$^2$-RAG)}, a two-stage framework that separates \emph{coverage} from \emph{commitment}. PAR$^2$-RAG first performs breadth-first anchoring to build a high-recall evidence frontier, then applies depth-first refinement with evidence sufficiency control in an iterative loop. Across four MHQA benchmarks, PAR$^2$-RAG consistently outperforms existing state-of-the-art baselines, compared with IRCoT, PAR$^2$-RAG achieves up to \textbf{23.5\%} higher accuracy, with retrieval gains of up to \textbf{10.5\%} in NDCG.

Webscraper: Leverage Multimodal Large Language Models for Index-Content Web Scraping

✇cs.AI, q-bio.NC updates on arXiv.org
By: Guan-Lun Huang · Yuh-Jzer Joung
1 April 2026 at 12:00
arXiv:2603.29161v1 Announce Type: new Abstract: Modern web scraping struggles with dynamic, interactive websites that require more than static HTML parsing. Current methods are often brittle and require manual customization for each site. To address this, we introduce Webscraper, a framework designed to handle the challenges of modern, dynamic web applications. It leverages a Multimodal Large Language Model (MLLM) to autonomously navigate interactive interfaces, invoke specialized tools, and perform structured data extraction in environments where traditional scrapers are ineffective. Webscraper utilizes a structured five-stage prompting procedure and a set of custom-built tools to navigate and extract data from websites following the common ``index-and-content'' architecture. Our experiments, conducted on six news websites, demonstrate that the full Webscraper framework, equipped with both our guiding prompt and specialized tools, achieves a significant improvement in extraction accuracy over the baseline agent Anthropic's Computer Use. We also applied the framework to e-commerce platforms to validate its generalizability.

Xuanwu: Evolving General Multimodal Models into an Industrial-Grade Foundation for Content Ecosystems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Zhiqian Zhang · Xu Zhao · Xiaoqing Xu · Guangdong Liang · Weijia Wang · Xiaolei Lv · Bo Li · Jun Gao
1 April 2026 at 12:00
arXiv:2603.29211v1 Announce Type: new Abstract: In recent years, multimodal large models have continued to improve on general benchmarks. However, in real-world content moderation and adversarial settings, mainstream models still suffer from degraded generalization and catastrophic forgetting because of limited fine-grained visual perception and insufficient modeling of long-tail noise. In this paper, we present Xuanwu VL-2B as a case study of how general multimodal models can be developed into an industrial-grade foundation model for content ecosystems. The model adopts a compact InternViT-300M + MLP + Qwen3 1.7B architecture, balancing fine-grained visual perception, language-semantic alignment, and deployment cost within an approximately 2B-parameter budget. To balance business specialization with the retention of general capabilities, we developed a data iteration and curation mechanism and trained the model through a progressive three-stage pipeline: pre-training, mid-training, and post-training. Ablation studies and offline business evaluations show that Xuanwu VL-2B achieves an average score of 67.90 across seven OpenCompass multimodal metrics (vs. 64.27 for InternVL 3.5 2B), an average recall of 94.38% over seven independent business moderation tasks, and a weighted overall recall of 82.82% on policy-violating text in challenging adversarial OCR scenarios, outperforming Gemini-2.5-Pro (76.72%). These results show that, under a limited parameter budget, Xuanwu VL-2B achieves a practical balance among business alignment, visual perception, general capability retention, and deployment cost.

ATP-Bench: Towards Agentic Tool Planning for MLLM Interleaved Generation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yinuo Liu · Zi Qian · Heng Zhou · Jiahao Zhang · Yajie Zhang · Zhihang Li · Mengyu Zhou · Erchao Zhao · Xiaoxi Jiang · Guanjun Jiang
1 April 2026 at 12:00
arXiv:2603.29902v1 Announce Type: new Abstract: Interleaved text-and-image generation represents a significant frontier for Multimodal Large Language Models (MLLMs), offering a more intuitive way to convey complex information. Current paradigms rely on either image generation or retrieval augmentation, yet they typically treat the two as mutually exclusive paths, failing to unify factuality with creativity. We argue that the next milestone in this field is Agentic Tool Planning, where the model serves as a central controller that autonomously determines when, where, and which tools to invoke to produce interleaved responses for visual-critical queries. To systematically evaluate this paradigm, we introduce ATP-Bench, a novel benchmark comprising 7,702 QA pairs (including 1,592 VQA pairs) across eight categories and 25 visual-critical intents, featuring human-verified queries and ground truths. Furthermore, to evaluate agentic planning independent of end-to-end execution and changing tool backends, we propose a Multi-Agent MLLM-as-a-Judge (MAM) system. MAM evaluates tool-call precision, identifies missed opportunities for tool use, and assesses overall response quality without requiring ground-truth references. Our extensive experiments on 10 state-of-the-art MLLMs reveal that models struggle with coherent interleaved planning and exhibit significant variations in tool-use behavior, highlighting substantial room for improvement and providing actionable guidance for advancing interleaved generation. Dataset and code are available at https://github.com/Qwen-Applications/ATP-Bench.
❌