Low-resource multilingual OCR faces a dual challenge: complex script structures and severe data scarcity.In such settings, existing OCR models often struggle, as coarse visual representations combined with weak linguistic priors lead to frequent errors among visually similar characters.To address this, we present BASA (Beyond Atomic Sub-character Alignment), a OCR framework built upon highresolution visual and language backbones with a novel glyph-aware interface.The core technical contribution is the Glyph-Aware Finegrained Adapter (GAFA).Unlike standard linear projectors, GAFA employs learnable glyph prototypes to actively align sub-character structural primitives (e.g., strokes and radicals) with visual features, explicitly resolving topological ambiguities during vision-language alignment.To complement this, we introduce a twostage curriculum learning strategy supported by a Glyph-Aware Reverse Synthesis pipeline, which generates large-scale multilingual training corpora with automatic, zero-cost component labels.Furthermore, we construct BASA-Bench, a representative benchmark spanning 11 languages with diverse script structures and 23 authentic scenarios.Experiments demonstrate that BASA achieves consistent improvements over strong OCR baselines, particularly on scripts with complex compositions.Our model and benchmark will be available at https://github.com/NcutLLM/BASA.

Large Language Models(LLMs) have brought significant transformations to various aspects of human life and productivity. However, the heavy reliance on vast amounts of data in developing these models has resulted in a notable disadvantage for low-resource languages, such as Nuosu and others, which lack large datasets. Moreover, many LLMs exhibit significant performance discrepancies between high-and lowresource languages, thereby restricting equitable access to technological advances for all linguistic communities. To address these challenges, this paper propose a low-resource multilingual large language model, termed VEEF-Multi-LLM, constructed through effective vocabulary expansion and parameter-efficient fine-tuning. We introduce a series of innovative methods to address challenges in low-resource languages. First, we adopt Byte-level Byte-Pair Encoding to expand the vocabulary for broader multilingual support. We separate input and output embedding weights to boost performance, and apply RoPE for long-context handling, as well as RMSNorm for efficient training. To generate high-quality supervised fine-tuning (SFT) data, we use self-training and selective translation, and refine the resulting dataset with the assistance of native speakers to ensure cultural and linguistic accuracy. Our model, VEEF-Multi-LLM-8B, is trained on 600 billion tokens across 50 natural and 16 programming languages. Experimental results show that the model excels in multilingual instruction-following tasks, particularly in translation, outperforming competing models in benchmarks such as XCOPA and XStoryCloze. Although it lags slightly behind English-centric models in some tasks (e.g., m-MMLU), it prioritizes safety, reliability, and inclusivity, making it valuable for diverse linguistic communities. We open-source our models on GitHub and Huggingface.

Multivariate time series anomaly detection plays a vital role in safety-critical domains such as industrial systems, finance, and cybersecurity. However, the scarcity of labeled anomalies poses significant challenges for learning robust normal patterns, often blurring the boundary between normal and abnormal behaviors. To address this challenge, we propose ADLM, an unsupervised adversarial framework that integrates a Language-Model-based Predictor for Time Series (LMPTS) with an autoencoder. To capture normal patterns under limited data, LMPTS repurposes a decoder-only pretrained language model as an autoregressive forecaster, leveraging its strong generative prior to capture temporal dependencies. To model complex cross-sensor dependencies, we incorporate graph structure learning into the framework. Furthermore, we introduce an adversarial training strategy to sharpen the model’s normal-pattern representations and amplify deviations indicative of anomalies. Experiments on six public datasets show that ADLM consistently outperforms state-of-the-art baselines and remains robust under severe data scarcity. By coupling decoder-only language models with an adversarial objective, ADLM offers a label-efficient, structure-aware solution to multivariate time series anomaly detection.

Text-Centric Visual Question Answering (TEC-VQA) serves as a key benchmark for evaluating AI's ability to reason over text-rich visual scenes.However, most existing TEC-VQA datasets focus on high-resource languages and are susceptible to benchmark contamination due to overlap with pretraining corpora of large models.These limitations severely hinder progress in low-resource language scenarios and compromise the reliability of current evaluations.To address both the underrepresentation of low-resource languages and the contamination issue, we propose TVQACML, the first large-scale TEC-VQA benchmark for multilingual Chinese minority languages, constructed through a scalable, reproducible pipeline.It comprises 8,000 real-world images and 32,000 high-quality QA pairs across eight languages and 30 application scenarios.We conduct comprehensive benchmarking of open-source, closed-source, and text-centric MLLMs, revealing substantial performance gaps from human accuracy, especially in scenetext and document understanding tasks.Furthermore, instruction tuning with TVQACML yields consistent performance gains, in some cases surpassing leading closed models demonstrating the dataset's utility for model alignment.We also introduce a lightweight, extensible evaluation metric for robust multilingual, multi-format answer assessment.The code and dataset for TVQACML are available at https://github.com/Shajiu/TVQACML.

Generating and creating humanoid 3D models has received increasing attention recently due to its fundamental support for many high-level 3D applications. Although automatic 3D pose and shape reconstruction methods have achieved promising results, there are still some failure cases due to self-occlusions, viewpoint changes, and the complexity of human pose articulations. In this paper, we propose a novel way to leverage Large Language Models (LLMs) to interactively reconstruct human pose and shape based on a Skinned Multi-Person Linear (SMPL) model. We construct a mapping table to fine-tune an LLM, enabling it to understand user inputs better and output the positional information of joint points. Additionally, a simple neural network is adopted to regress the shape cues of the SMPL. We demonstrate a gallery of results of numerous poses and shapes. We validate our method via numerical evaluations, user studies, and comparisons to manually posed characters and previous work.

The evolution of blockchain technology across various areas has highlighted the importance of optimizing blockchain systems' performance, especially in fluctuating network bandwidth conditions. We observe that the performance of blockchain systems exhibits variations, and the optimal parameter configuration shifts accordingly when changes in network bandwidth occur. Current methods in blockchain optimization require establishing fixed mappings between various environments and their optimal parameters. However, this process exhibits poor sample efficiency and lacks the ability for fast adaptation to novel bandwidth environments. In this paper, we propose MetaTune, a meta-Reinforcement-Learning (meta-RL)-based dynamic tuning method for blockchain systems. MetaTune can quickly adapt to unknown bandwidth changes and automatically configure optimized parameters. Through empirical evaluations of a real-world blockchain system, ChainMaker, we demonstrate that MetaTune significantly reduces the training samples needed for generalization across different bandwidth environments compared to non-adaptive methods. Our findings suggest that MetaTune offers a promising approach for efficiently optimizing blockchain systems in dynamic network environments.

ABSTRACT Blockchain systems, characterized by decentralization, irreversibility, and traceability, have attracted widespread adoption across security‐critical domains. However, the discrepancy between claimed and observed performance metrics poses significant challenges for system selection and reliability assurance. Particularly, in multi‐layer blockchain architectures with complex inter‐node interactions, pinpointing abnormal nodes or malfunctioning stages remains a non‐trivial task. While macroscopic metrics, such as transactions per second (TPS) and average latency are essential for measuring system capacity, they lack the granularity to capture fine‐grained operational anomalies. To complement these metrics and provide a new diagnostic dimension, we propose a novel metric system grounded in the spatio‐temporal transition of transaction states, introducing three complementary indicators: overall spatio‐temporal transition cost, spatial transition cost, and temporal transition cost. These metrics characterize system‐wide behavior and enable anomaly detection at both the node and stage levels. We further develop a log‐driven analysis framework that leverages these metrics for anomaly localization and root cause inference. Our system is implemented atop ChainMaker, deployed over 16 containerized nodes using Docker and Kubernetes. Experimental results demonstrate that our proposed metrics exhibit strong stability, sensitivity, and precision in detecting abnormal behaviors under varied execution conditions. These results validate the effectiveness of our methodology in providing fine‐grained insights into the performance reliability of blockchain systems.

Large language models (LLMs), such as ChatGPT and LLama, have shown remarkable capability in a wide range of natural language tasks. However, the current LLMs are mainly concentrated in resource-rich languages, such as English and Chinese. For low-resource language such as Tibetan, research and applications related to LLMs are still in their infancy. To address the existing gap, we present a method to enhance LLaMA with the ability to understand and generate Tibetan text, as well as to follow instructions. This is achieved by creating large-scale unsupervised pre-training and supervised fine-tuning datasets, which mitigate the limited availability of Tibetan data. Additionally, we expand LLaMA’s vocabulary by incorporating Tibetan tokens through Unigram tokenization, thereby improving both its encoding efficiency and semantic understanding of Tibetan. Furthermore, we conduct secondary pre-training and fine-tune the model using the constructed datasets, thereby enhancing its capability to interpret and execute instructions effectively. To verify the effectiveness of the model, we establish ten evaluation benchmarks for Tibetan. The experimental results indicate that the proposed model significantly enhances the LLaMA’s proficiency in understanding and generating Tibetan content. To promote further research, we release our model and inference resources at https://github.com/Shajiu/Tibetan-LLaMA-2 .