Qian Jiang

Yicheng Zou, Dongsheng Zhu, Lin Zhu et al.

We introduce Intern-S1-Pro, the first one-trillion-parameter scientific multimodal foundation model. Scaling to this unprecedented size, the model delivers a comprehensive enhancement across both general and scientific domains. Beyond stronger reasoning and image-text understanding capabilities, its intelligence is augmented with advanced agent capabilities. Simultaneously, its scientific expertise has been vastly expanded to master over 100 specialized tasks across critical science fields, including chemistry, materials, life sciences, and earth sciences. Achieving this massive scale is made possible by the robust infrastructure support of XTuner and LMDeploy, which facilitates highly efficient Reinforcement Learning (RL) training at the 1-trillion parameter level while ensuring strict precision consistency between training and inference. By seamlessly integrating these advancements, Intern-S1-Pro further fortifies the fusion of general and specialized intelligence, working as a Specializable Generalist, demonstrating its position in the top tier of open-source models for general capabilities, while outperforming proprietary models in the depth of specialized scientific tasks.

Qian Jiang, Changyou Chen, Han Zhao et al.

Contrastive loss has been increasingly used in learning representations from multiple modalities. In the limit, the nature of the contrastive loss encourages modalities to exactly match each other in the latent space. Yet it remains an open question how the modality alignment affects the downstream task performance. In this paper, based on an information-theoretic argument, we first prove that exact modality alignment is sub-optimal in general for downstream prediction tasks. Hence we advocate that the key of better performance lies in meaningful latent modality structures instead of perfect modality alignment. To this end, we propose three general approaches to construct latent modality structures. Specifically, we design 1) a deep feature separation loss for intra-modality regularization; 2) a Brownian-bridge loss for inter-modality regularization; and 3) a geometric consistency loss for both intra- and inter-modality regularization. Extensive experiments are conducted on two popular multi-modal representation learning frameworks: the CLIP-based two-tower model and the ALBEF-based fusion model. We test our model on a variety of tasks including zero/few-shot image classification, image-text retrieval, visual question answering, visual reasoning, and visual entailment. Our method achieves consistent improvements over existing methods, demonstrating the effectiveness and generalizability of our proposed approach on latent modality structure regularization.

Dongjian Yu, Weiqing Min, Qian Jiang et al.

Accurate estimation of food nutrition plays a vital role in promoting healthy dietary habits and personalized diet management. Most existing food datasets primarily focus on Western cuisines and lack sufficient coverage of Chinese dishes, which restricts accurate nutritional estimation for Chinese meals. Moreover, many state-of-the-art nutrition prediction methods rely on depth sensors, restricting their applicability in daily scenarios. To address these limitations, we introduce OmniFood8K, a comprehensive multimodal dataset comprising 8,036 food samples, each with detailed nutritional annotations and multi-view images. In addition, to enhance models' capability in nutritional prediction, we construct NutritionSynth-115K, a large-scale synthetic dataset that introduces compositional variations while preserving precise nutritional labels. Moreover, we propose an end-to-end framework for nutritional prediction from a single RGB image. First, we predict a depth map from a single RGB image and design the Scale-Shift Residual Adapter (SSRA) to refine it for global scale consistency and local structural preservation. Second, we propose the Frequency-Aligned Fusion Module (FAFM) to hierarchically align and fuse RGB and depth features in the frequency domain. Finally, we design a Mask-based Prediction Head (MPH) to emphasize key ingredient regions via dynamic channel selection for more accurate prediction. Extensive experiments on multiple datasets demonstrate the superiority of our method over existing approaches. Project homepage: https://yudongjian.github.io/OmniFood8K-food/

Fengxian Ji, Jingpu Yang, Zirui Song et al.

Despite the rapid progress of large vision-language models (LVLMs), fine-grained, state-conditioned GUI interaction remains challenging. Current evaluations offer limited coverage, imprecise target-state definitions, and an overreliance on final-task success, obscuring where and why agents fail. To address this gap, we introduce \textbf{FineState-Bench}, a benchmark that evaluates whether an agent can correctly ground an instruction to the intended UI control and reach the exact target state. FineState-Bench comprises 2,209 instances across desktop, web, and mobile platforms, spanning four interaction families and 23 UI component types, with each instance explicitly specifying an exact target state for fine-grained state setting. We further propose \textit{FineState-Metrics}, a four-stage diagnostic pipeline with stage-wise success rates: Localization Success Rate (SR@Loc), Interaction Success Rate (SR@Int), Exact State Success Rate at Locate (ES-SR@Loc), and Exact State Success Rate at Interact (ES-SR@Int), and a plug-and-play \textit{Visual Diagnostic Assistant} (VDA) that generates a Description and a bounding-box Localization Hint to diagnose visual grounding reason via controlled w/ vs.\ w/o comparisons. On FineState-Bench, exact goal-state success remains low: ES-SR@Int peaks at 32.8\% on Web and 22.8\% on average across platforms. With VDA localization hints, Gemini-2.5-Flash gains +14.9 ES-SR@Int points, suggesting substantial headroom from improved visual grounding, yet overall accuracy is still insufficient for reliable fine-grained state-conditioned interaction \href{https://github.com/FengxianJi/FineState-Bench}{Github.}

Mingxuan Cui, Jingpu Yang, Fengxian Ji et al.

Faithful text rendering remains a persistent weakness of large text-to-image generative models, as it requires both semantic instruction following and fine-grained glyph-level structure. Prior methods often improve this ability through architecture-specific modules or encoder modifications, which complicate deployment across foundation models. We study text rendering as a post-training preference-alignment problem and propose TextAlign, a non-invasive framework that keeps the generator architecture unchanged. The key component is a hierarchical vision-language model (VLM)-based reward that decomposes rendering errors into global, word, and glyph levels, then converts binary defect judgments into a scalar preference signal. The resulting signal supports both Group Relative Policy Optimization (GRPO) and Direct Preference Optimization (DPO). Experiments on FLUX.1-dev and Z-Image-Turbo show consistent gains in OCR-based text accuracy without degrading general generation quality. Compared with strong foundation and text-rendering baselines, including SD3.5, Qwen-Image, AnyText, and TextDiffuser, these results indicate that reward design offers a scalable alternative to model redesign for improving text rendering.

Qian Jiang, Liping Sun

Scientific forecasting typically relies on direct state prediction, an approach that grows brittle under data scarcity, extended horizons, non-stationary dynamics, or high-dimensional complexity. While raw state trajectories are highly sensitive in these regimes, underlying local evolution rules often exhibit robust reusability. We introduce mechanism learning, a framework that forecasts future states by estimating the currently active local mechanism. Our method compresses local spatiotemporal fragments into mechanism descriptors, forming a data-driven, structured mechanism space where proximity reflects similar local evolution rules. To ground these estimates in observed data, we utilize prototype anchors, a set of representative mechanisms that sparsely cover the space of local rules. We evaluate this approach on Burgers dynamics, WeatherBench2, and Lorenz96. Empirically, the learned mechanism spaces resist collapse and maintain strong local consistency. Compared to direct prediction and other models including FNO, NODE, LSTM, and reservoir-family methods, our framework demonstrates predictive gains in fragile regimes: it significantly improves switching stability in Burgers dynamics and achieves state-of-the-art performance both under the scarce-data fixed-horizon WeatherBench2 protocol and in intermediate-complexity Lorenz96. Ablation studies and drift diagnostics confirm that these improvements are driven by finite prototype anchoring rather than sheer latent capacity. Together, these results establish mechanism learning as a principled, robust alternative to direct state prediction in forecasting complex systems.

Fengxian Ji, Jingpu Yang, Zirui Song et al.

With the rapid advancement of generative artificial intelligence technology, Graphical User Interface (GUI) agents have demonstrated tremendous potential for autonomously managing daily tasks through natural language instructions. However, current evaluation frameworks for GUI agents suffer from fundamental flaws: existing benchmarks overly focus on coarse-grained task completion while neglecting fine-grained control capabilities crucial for real-world applications. To address this, we introduce FineState-Bench, the first evaluation and diagnostic standard for fine-grained GUI proxy operations, designed to quantify fine-grained control. This multi-platform (desktop, Web, mobile) framework includes 2257 task benchmarks in four components and uses a four-phase indicator for comprehensive perception-to-control assessment. To analyze perception and positioning for refined operations, we developed the plug-and-play Visual Diagnostic Assistant (VDA), enabling the first quantitative decoupling analysis of these capabilities. Experimental results on our benchmark show that the most advanced models achieve only 32.8% fine-grained interaction accuracy. Using our VDA in controlled experiments, quantifying the impact of visual capabilities, we showed that ideal visual localization boosts Gemini-2.5-Flash's success rate by 14.9\%. Our diagnostic framework confirms for the first time that the primary bottleneck for current GUI proxies is basic visual positioning capability.All resources are fully open-source. github: https://github.com/AnonymousThewarehouse/FineState-Bench huggingface: https://huggingface.co/datasets/Willtime2006/Static-FineBench

Huimin Zeng, Zhenrui Yue, Qian Jiang et al.

Federated Recommendation (FR) emerges as a novel paradigm that enables privacy-preserving recommendations. However, traditional FR systems usually represent users/items with discrete identities (IDs), suffering from performance degradation due to the data sparsity and heterogeneity in FR. On the other hand, Large Language Models (LLMs) as recommenders have proven effective across various recommendation scenarios. Yet, LLM-based recommenders encounter challenges such as low inference efficiency and potential hallucination, compromising their performance in real-world scenarios. To this end, we propose GPT-FedRec, a federated recommendation framework leveraging ChatGPT and a novel hybrid Retrieval Augmented Generation (RAG) mechanism. GPT-FedRec is a two-stage solution. The first stage is a hybrid retrieval process, mining ID-based user patterns and text-based item features. Next, the retrieved results are converted into text prompts and fed into GPT for re-ranking. Our proposed hybrid retrieval mechanism and LLM-based re-rank aims to extract generalized features from data and exploit pretrained knowledge within LLM, overcoming data sparsity and heterogeneity in FR. In addition, the RAG approach also prevents LLM hallucination, improving the recommendation performance for real-world users. Experimental results on diverse benchmark datasets demonstrate the superior performance of GPT-FedRec against state-of-the-art baseline methods.

Zirui Song, Guangxian Ouyang, Mingzhe Li et al.

Large Vision-Language Models (LVLMs) have recently advanced robotic manipulation by leveraging vision for scene perception and language for instruction following. However, existing methods rely heavily on costly human-annotated training datasets, which limits their generalization and causes them to struggle in out-of-domain (OOD) scenarios, reducing real-world adaptability. To address these challenges, we propose ManipLVM-R1, a novel reinforcement learning framework that replaces traditional supervision with Reinforcement Learning using Verifiable Rewards (RLVR). By directly optimizing for task-aligned outcomes, our method enhances generalization and physical reasoning while removing the dependence on costly annotations. Specifically, we design two rule-based reward functions targeting key robotic manipulation subtasks: an Affordance Perception Reward to enhance localization of interaction regions, and a Trajectory Match Reward to ensure the physical plausibility of action paths. These rewards provide immediate feedback and impose spatial-logical constraints, encouraging the model to go beyond shallow pattern matching and instead learn deeper, more systematic reasoning about physical interactions.

Zirui Song, Qian Jiang, Mingxuan Cui et al.

The rise of Large Audio Language Models (LAMs) brings both potential and risks, as their audio outputs may contain harmful or unethical content. However, current research lacks a systematic, quantitative evaluation of LAM safety especially against jailbreak attacks, which are challenging due to the temporal and semantic nature of speech. To bridge this gap, we introduce AJailBench, the first benchmark specifically designed to evaluate jailbreak vulnerabilities in LAMs. We begin by constructing AJailBench-Base, a dataset of 1,495 adversarial audio prompts spanning 10 policy-violating categories, converted from textual jailbreak attacks using realistic text to speech synthesis. Using this dataset, we evaluate several state-of-the-art LAMs and reveal that none exhibit consistent robustness across attacks. To further strengthen jailbreak testing and simulate more realistic attack conditions, we propose a method to generate dynamic adversarial variants. Our Audio Perturbation Toolkit (APT) applies targeted distortions across time, frequency, and amplitude domains. To preserve the original jailbreak intent, we enforce a semantic consistency constraint and employ Bayesian optimization to efficiently search for perturbations that are both subtle and highly effective. This results in AJailBench-APT, an extended dataset of optimized adversarial audio samples. Our findings demonstrate that even small, semantically preserved perturbations can significantly reduce the safety performance of leading LAMs, underscoring the need for more robust and semantically aware defense mechanisms.

Ren Ma, Jiantao Qiu, Chao Xu et al.

While web data quality is crucial for large language models, most curation efforts focus on filtering and deduplication,treating HTML-to-text extraction as a fixed pre-processing step. Existing web corpora rely on heuristic-based extractors like Trafilatura, which struggle to preserve document structure and frequently corrupt structured elements such as formulas, codes, and tables. We hypothesize that improving extraction quality can be as impactful as aggressive filtering strategies for downstream performance. We introduce MinerU-HTML, a novel extraction pipeline that reformulates content extraction as a sequence labeling problem solved by a 0.6B-parameter language model. Unlike text-density heuristics, MinerU-HTML leverages semantic understanding and employs a two-stage formatting pipeline that explicitly categorizes semantic elements before converting to Markdown. Crucially, its model-based approach is inherently scalable, whereas heuristic methods offer limited improvement pathways. On MainWebBench, our benchmark of 7,887 annotated web pages, MinerU-HTML achieves 81.8\% ROUGE-N F1 compared to Trafilatura's 63.6\%, with exceptional structured element preservation (90.9\% for code blocks, 94.0\% for formulas). Using MinerU-HTML, we construct AICC (AI-ready Common Crawl), a 7.3-trillion token multilingual corpus from two Common Crawl snapshots. In controlled pretraining experiments where AICC and Trafilatura-extracted TfCC undergo identical filtering, models trained on AICC (62B tokens) achieve 50.8\% average accuracy across 13 benchmarks, outperforming TfCC by 1.08pp-providing direct evidence that extraction quality significantly impacts model capabilities. AICC also surpasses RefinedWeb and FineWeb on key benchmarks. We publicly release MainWebBench, MinerU-HTML, and AICC, demonstrating that HTML extraction is a critical, often underestimated component of web corpus construction.

Qian Jiang, Qianqian Wang, Xin Jin et al.

Remote sensing images are becoming increasingly widespread in military, earth resource exploration. Because of the limitation of a single sensor, we can obtain high spatial resolution grayscale panchromatic (PAN) images and low spatial resolution color multispectral (MS) images. Therefore, an important issue is to obtain a color image with high spatial resolution when there is only a PAN image at the input. The existing methods improve spatial resolution using super-resolution (SR) technology and spectral recovery using colorization technology. However, the SR technique cannot improve the spectral resolution, and the colorization technique cannot improve the spatial resolution. Moreover, the pansharpening method needs two registered inputs and can not achieve SR. As a result, an integrated approach is expected. To solve the above problems, we designed a novel multi-function model (MFmamba) to realize the tasks of SR, spectral recovery, joint SR and spectral recovery through three different inputs. Firstly, MFmamba utilizes UNet++ as the backbone, and a Mamba Upsample Block (MUB) is combined with UNet++. Secondly, a Dual Pool Attention (DPA) is designed to replace the skip connection in UNet++. Finally, a Multi-scale Hybrid Cross Block (MHCB) is proposed for initial feature extraction. Many experiments show that MFmamba is competitive in evaluation metrics and visual results and performs well in the three tasks when only the input PAN image is used.

Qian Jiang, Xiaofan Zhang, Deming Chen et al.

In hardware-aware Differentiable Neural Architecture Search (DNAS), it is challenging to compute gradients of hardware metrics to perform architecture search. Existing works rely on linear approximations with limited support to customized hardware accelerators. In this work, we propose End-to-end Hardware-aware DNAS (EH-DNAS), a seamless integration of end-to-end hardware benchmarking, and fully automated DNAS to deliver hardware-efficient deep neural networks on various platforms, including Edge GPUs, Edge TPUs, Mobile CPUs, and customized accelerators. Given a desired hardware platform, we propose to learn a differentiable model predicting the end-to-end hardware performance of neural network architectures for DNAS. We also introduce E2E-Perf, an end-to-end hardware benchmarking tool for customized accelerators. Experiments on CIFAR10 and ImageNet show that EH-DNAS improves the hardware performance by an average of $1.4\times$ on customized accelerators and $1.6\times$ on existing hardware processors while maintaining the classification accuracy.

Gaojian Wang, Qian Jiang, Xin Jin et al.

As the remarkable development of facial manipulation technologies is accompanied by severe security concerns, face forgery detection has become a recent research hotspot. Most existing detection methods train a binary classifier under global supervision to judge real or fake. However, advanced manipulations only perform small-scale tampering, posing challenges to comprehensively capture subtle and local forgery artifacts, especially in high compression settings and cross-dataset scenarios. To address such limitations, we propose a novel framework named Multi-modal Contrastive Classification by Locally Correlated Representations(MC-LCR), for effective face forgery detection. Instead of specific appearance features, our MC-LCR aims to amplify implicit local discrepancies between authentic and forged faces from both spatial and frequency domains. Specifically, we design the shallow style representation block that measures the pairwise correlation of shallow feature maps, which encodes local style information to extract more discriminative features in the spatial domain. Moreover, we make a key observation that subtle forgery artifacts can be further exposed in the patch-wise phase and amplitude spectrum and exhibit different clues. According to the complementarity of amplitude and phase information, we develop a patch-wise amplitude and phase dual attention module to capture locally correlated inconsistencies with each other in the frequency domain. Besides the above two modules, we further introduce the collaboration of supervised contrastive loss with cross-entropy loss. It helps the network learn more discriminative and generalized representations. Through extensive experiments and comprehensive studies, we achieve state-of-the-art performance and demonstrate the robustness and generalization of our method.

Gaojian Wang, Qian Jiang, Xin Jin et al.

The internet is filled with fake face images and videos synthesized by deep generative models. These realistic DeepFakes pose a challenge to determine the authenticity of multimedia content. As countermeasures, artifact-based detection methods suffer from insufficiently fine-grained features that lead to limited detection performance. DNN-based detection methods are not efficient enough, given that a DeepFake can be created easily by mobile apps and DNN-based models require high computational resources. For the first time, we show that DeepFake faces have fewer feature points than real ones, especially in certain facial regions. Inspired by feature point detector-descriptors to extract discriminative features at the pixel level, we propose the Fused Facial Region_Feature Descriptor (FFR_FD) for effective and fast DeepFake detection. FFR_FD is only a vector extracted from the face, and it can be constructed from any feature point detector-descriptors. We train a random forest classifier with FFR_FD and conduct extensive experiments on six large-scale DeepFake datasets, whose results demonstrate that our method is superior to most state of the art DNN-based models.