Haonan Li

Neha Sengupta, Sunil Kumar Sahu, Bokang Jia et al. · berkeley

We introduce Jais and Jais-chat, new state-of-the-art Arabic-centric foundation and instruction-tuned open generative large language models (LLMs). The models are based on the GPT-3 decoder-only architecture and are pretrained on a mixture of Arabic and English texts, including source code in various programming languages. With 13 billion parameters, they demonstrate better knowledge and reasoning capabilities in Arabic than any existing open Arabic and multilingual models by a sizable margin, based on extensive evaluation. Moreover, the models are competitive in English compared to English-centric open models of similar size, despite being trained on much less English data. We provide a detailed description of the training, the tuning, the safety alignment, and the evaluation of the models. We release two open versions of the model -- the foundation Jais model, and an instruction-tuned Jais-chat variant -- with the aim of promoting research on Arabic LLMs. Available at https://huggingface.co/inception-mbzuai/jais-13b-chat

Yuxia Wang, Haonan Li, Xudong Han et al.

With the rapid evolution of large language models (LLMs), new and hard-to-predict harmful capabilities are emerging. This requires developers to be able to identify risks through the evaluation of "dangerous capabilities" in order to responsibly deploy LLMs. In this work, we collect the first open-source dataset to evaluate safeguards in LLMs, and deploy safer open-source LLMs at a low cost. Our dataset is curated and filtered to consist only of instructions that responsible language models should not follow. We annotate and assess the responses of six popular LLMs to these instructions. Based on our annotation, we proceed to train several BERT-like classifiers, and find that these small classifiers can achieve results that are comparable with GPT-4 on automatic safety evaluation. Warning: this paper contains example data that may be offensive, harmful, or biased.

Shuai Fan, Chen Lin, Haonan Li et al.

Most existing pre-trained language representation models (PLMs) are sub-optimal in sentiment analysis tasks, as they capture the sentiment information from word-level while under-considering sentence-level information. In this paper, we propose SentiWSP, a novel Sentiment-aware pre-trained language model with combined Word-level and Sentence-level Pre-training tasks. The word level pre-training task detects replaced sentiment words, via a generator-discriminator framework, to enhance the PLM's knowledge about sentiment words. The sentence level pre-training task further strengthens the discriminator via a contrastive learning framework, with similar sentences as negative samples, to encode sentiments in a sentence. Extensive experimental results show that SentiWSP achieves new state-of-the-art performance on various sentence-level and aspect-level sentiment classification benchmarks. We have made our code and model publicly available at https://github.com/XMUDM/SentiWSP.

Haonan Li, Yixuan Zhang, Fajri Koto et al.

As the capabilities of large language models (LLMs) continue to advance, evaluating their performance becomes increasingly crucial and challenging. This paper aims to bridge this gap by introducing CMMLU, a comprehensive Chinese benchmark that covers various subjects, including natural science, social sciences, engineering, and humanities. We conduct a thorough evaluation of 18 advanced multilingual- and Chinese-oriented LLMs, assessing their performance across different subjects and settings. The results reveal that most existing LLMs struggle to achieve an average accuracy of 50%, even when provided with in-context examples and chain-of-thought prompts, whereas the random baseline stands at 25%. This highlights significant room for improvement in LLMs. Additionally, we conduct extensive experiments to identify factors impacting the models' performance and propose directions for enhancing LLMs. CMMLU fills the gap in evaluating the knowledge and reasoning capabilities of large language models within the Chinese context.

Saeed Almheiri, Bilal Elbouardi, Salsabila Zahirah Pranida et al.

Idiomatic expressions pose a major challenge for multilingual NLP because their meanings shift between figurative and literal usage, often requiring context for accurate interpretation. Prior work has focused on high-resource languages typically evaluates isolated idiom-meaning questions, overlooking realistic discourse. We introduce MIDI, a multilingual idiom dataset spanning 3 high-, 3 medium-, and 12 low-resource languages, curated by native speakers. Unlike previous datasets, MIDI provides idioms embedded in both sentence-level and conversational contexts, capturing both literal and figurative readings. Benchmarking state-of-the-art models shows that idiom comprehension degrades in low-resource languages and that, in all resource tiers, literal interpretations are substantially harder than figurative ones. Conversational context improves performance but does not eliminate these disparities. Through controlled tests and interventions on hidden representations, we further separate memorization from reasoning, exposing core limitations of current models.

Haonan Li, Yu Hao, Yizhuo Zhai et al.

Static analysis is a widely used technique in software engineering for identifying and mitigating bugs. However, a significant hurdle lies in achieving a delicate balance between precision and scalability. Large Language Models (LLMs) offer a promising alternative, as recent advances demonstrate remarkable capabilities in comprehending, generating, and even debugging code. Yet, the logic of bugs can be complex and require sophisticated reasoning and a large analysis scope spanning multiple functions. Therefore, at this point, LLMs are better used in an assistive role to complement static analysis. In this paper, we take a deep dive into the open space of LLM-assisted static analysis, using use-before-initialization (UBI) bugs as a case study. To this end, we develop LLift, a fully automated framework that interfaces with both a static analysis tool and an LLM. By carefully designing the framework and the prompts, we are able to overcome a number of challenges, including bug-specific modeling, the large problem scope, the non-deterministic nature of LLMs, etc. Tested in a real-world scenario analyzing nearly a thousand potential UBI bugs produced by static analysis, LLift demonstrates a potent capability, showcasing a reasonable precision (50%) and appearing to have no missing bugs. It even identified 13 previously unknown UBI bugs in the Linux kernel. This research paves the way for new opportunities and methodologies in using LLMs for bug discovery in extensive, real-world datasets.

Fajri Koto, Nurul Aisyah, Haonan Li et al.

Although large language models (LLMs) are often pre-trained on large-scale multilingual texts, their reasoning abilities and real-world knowledge are mainly evaluated based on English datasets. Assessing LLM capabilities beyond English is increasingly vital but hindered due to the lack of suitable datasets. In this work, we introduce IndoMMLU, the first multi-task language understanding benchmark for Indonesian culture and languages, which consists of questions from primary school to university entrance exams in Indonesia. By employing professional teachers, we obtain 14,981 questions across 64 tasks and education levels, with 46% of the questions focusing on assessing proficiency in the Indonesian language and knowledge of nine local languages and cultures in Indonesia. Our empirical evaluations show that GPT-3.5 only manages to pass the Indonesian primary school level, with limited knowledge of local Indonesian languages and culture. Other smaller models such as BLOOMZ and Falcon perform at even lower levels.

Ya-Qi Yu, Hao Wang, Fangyu Hong et al.

While Reinforcement Learning with Verifiable Rewards (RLVR) is effective for deterministically checkable tasks, many vision-language tasks are partially verifiable, demanding multi-criteria supervision (e.g., perceptual details, reasoning steps, and constraints). Rubrics provide a natural interface for this fine-grained supervision, but their effectiveness depends on the execution accuracy during online RL. We propose Reinforcement Learning with Robust Rubric Rewards ($\text{RLR}^3$), extending RLVR from task-level verification to criterion-level verification. $\text{RLR}^3$ routes instance-specific rubrics through two execution paths: an LLM-as-an-extractor paired with a deterministic verifier, or an LLM-as-a-Judge for non-verifiable criteria. To ensure faithful scoring, $\text{RLR}^3$ introduce a minimal exposure strategy that masks ground truths from extractors and images from judges. Furthermore, $\text{RLR}^3$ employs hierarchical aggregation to prioritize essential criteria over additional criteria, and mitigates score saturation within rollout groups. Evaluated on Qwen3-VL-30B-A3B across 15 benchmarks, $\text{RLR}^3$ consistently outperforms RLVR, yielding a 4.7-point improvement over the base model and exceeding the official instruct-to-thinking model gap. Controlled audits confirm our deterministic verification and minimal exposure significantly reduce exploitable false positives.

Stella Biderman, Hailey Schoelkopf, Lintang Sutawika et al. · cmu

Effective evaluation of language models remains an open challenge in NLP. Researchers and engineers face methodological issues such as the sensitivity of models to evaluation setup, difficulty of proper comparisons across methods, and the lack of reproducibility and transparency. In this paper we draw on three years of experience in evaluating large language models to provide guidance and lessons for researchers. First, we provide an overview of common challenges faced in language model evaluation. Second, we delineate best practices for addressing or lessening the impact of these challenges on research. Third, we present the Language Model Evaluation Harness (lm-eval): an open source library for independent, reproducible, and extensible evaluation of language models that seeks to address these issues. We describe the features of the library as well as case studies in which the library has been used to alleviate these methodological concerns.

Yixuan Zhang, Haonan Li

Large language models (LLMs) have showcased remarkable capabilities in understanding and generating language. However, their ability in comprehending ancient languages, particularly ancient Chinese, remains largely unexplored. To bridge this gap, we present ACLUE, an evaluation benchmark designed to assess the capability of language models in comprehending ancient Chinese. ACLUE consists of 15 tasks cover a range of skills, spanning phonetic, lexical, syntactic, semantic, inference and knowledge. Through the evaluation of eight state-of-the-art LLMs, we observed a noticeable disparity in their performance between modern Chinese and ancient Chinese. Among the assessed models, ChatGLM2 demonstrates the most remarkable performance, achieving an average score of 37.4%. We have made our code and data public available.

Zhengzhong Liu, Aurick Qiao, Willie Neiswanger et al.

The recent surge in open-source Large Language Models (LLMs), such as LLaMA, Falcon, and Mistral, provides diverse options for AI practitioners and researchers. However, most LLMs have only released partial artifacts, such as the final model weights or inference code, and technical reports increasingly limit their scope to high-level design choices and surface statistics. These choices hinder progress in the field by degrading transparency into the training of LLMs and forcing teams to rediscover many details in the training process. We present LLM360, an initiative to fully open-source LLMs, which advocates for all training code and data, model checkpoints, and intermediate results to be made available to the community. The goal of LLM360 is to support open and collaborative AI research by making the end-to-end LLM training process transparent and reproducible by everyone. As a first step of LLM360, we release two 7B parameter LLMs pre-trained from scratch, Amber and CrystalCoder, including their training code, data, intermediate checkpoints, and analyses (at https://www.llm360.ai). We are committed to continually pushing the boundaries of LLMs through this open-source effort. More large-scale and stronger models are underway and will be released in the future.

Fajri Koto, Haonan Li, Sara Shatnawi et al.

The focus of language model evaluation has transitioned towards reasoning and knowledge-intensive tasks, driven by advancements in pretraining large models. While state-of-the-art models are partially trained on large Arabic texts, evaluating their performance in Arabic remains challenging due to the limited availability of relevant datasets. To bridge this gap, we present \datasetname{}, the first multi-task language understanding benchmark for the Arabic language, sourced from school exams across diverse educational levels in different countries spanning North Africa, the Levant, and the Gulf regions. Our data comprises 40 tasks and 14,575 multiple-choice questions in Modern Standard Arabic (MSA) and is carefully constructed by collaborating with native speakers in the region. Our comprehensive evaluations of 35 models reveal substantial room for improvement, particularly among the best open-source models. Notably, BLOOMZ, mT0, LLaMA2, and Falcon struggle to achieve a score of 50%, while even the top-performing Arabic-centric model only achieves a score of 62.3%.

Ya-Qi Yu, Fangyu Hong, Xiangyang Qu et al.

The effectiveness of Direct Preference Optimization (DPO) depends on preference data that reflect the quality differences that matter in multimodal tasks. Existing pipelines often rely on off-policy perturbations or coarse outcome-based signals, which are not well suited to fine-grained visual reasoning. We propose rDPO, a preference optimization framework based on instance-specific rubrics. For each image-instruction pair, we create a checklist-style rubric of essential and additional criteria to score responses from any possible policies. The instruction-rubric pool is built offline and reused during the construction of on-policy data. On public reward modeling benchmarks, rubric-based prompting massively improves a 30B-A3B judge and brings it close to GPT-5.4. On public downstream benchmarks, rubric-based filtering raises the macro average to 82.69, whereas outcome-based filtering drops it to 75.82 from 81.14. When evaluating scalability on a comprehensive benchmark, rDPO achieves 61.01, markedly outperforming the style-constrained baseline (52.36) and surpassing the 59.48 base model. Together, these results show that visual preference optimization benefits from combining on-policy data construction with instance-specific criterion-level feedback.

Yanan Wang, Renxi Wang, Yongxin Wang et al.

Large language models (LLMs) have been extensively studied for tasks like math competitions, complex coding, and scientific reasoning, yet their ability to accurately represent and simulate physical scenarios via code remains underexplored. We propose SimuScene, the first systematic study that trains and evaluates LLMs on simulating physical scenarios across five physics domains and 52 physical concepts. We build an automatic pipeline to collect data, with human verification to ensure quality. The final dataset contains 7,659 physical scenarios with 334 human-verified examples as the test set. We evaluated 10 contemporary LLMs and found that even the strongest model achieves only a 21.5% pass rate, demonstrating the difficulty of the task. Finally, we introduce a reinforcement learning pipeline with visual rewards that uses a vision-language model as a judge to train textual models. Experiments show that training with our data improves physical simulation via code while substantially enhancing general code generation performance.

Zhoujun Cheng, Shibo Hao, Tianyang Liu et al.

Reinforcement learning (RL) has emerged as a promising approach to improve large language model (LLM) reasoning, yet most open efforts focus narrowly on math and code, limiting our understanding of its broader applicability to general reasoning. A key challenge lies in the lack of reliable, scalable RL reward signals across diverse reasoning domains. We introduce Guru, a curated RL reasoning corpus of 92K verifiable examples spanning six reasoning domains--Math, Code, Science, Logic, Simulation, and Tabular--each built through domain-specific reward design, deduplication, and filtering to ensure reliability and effectiveness for RL training. Based on Guru, we systematically revisit established findings in RL for LLM reasoning and observe significant variation across domains. For example, while prior work suggests that RL primarily elicits existing knowledge from pretrained models, our results reveal a more nuanced pattern: domains frequently seen during pretraining (Math, Code, Science) easily benefit from cross-domain RL training, while domains with limited pretraining exposure (Logic, Simulation, and Tabular) require in-domain training to achieve meaningful performance gains, suggesting that RL is likely to facilitate genuine skill acquisition. Finally, we present Guru-7B and Guru-32B, two models that achieve state-of-the-art performance among open models RL-trained with publicly available data, outperforming best baselines by 7.9% and 6.7% on our 17-task evaluation suite across six reasoning domains. We also show that our models effectively improve the Pass@k performance of their base models, particularly on complex tasks less likely to appear in pretraining data. We release data, models, training and evaluation code to facilitate general-purpose reasoning at: https://github.com/LLM360/Reasoning360

Chengqian Gao, Haonan Li, Liu Liu et al.

The alignment of large language models (LLMs) often assumes that using more clean data yields better outcomes, overlooking the match between model capacity and example difficulty. Challenging this, we propose a new principle: Preference data vary in difficulty, and overly difficult examples hinder alignment, by exceeding the model's capacity. Through systematic experimentation, we validate this principle with three key findings: (1) preference examples vary in difficulty, as evidenced by consistent learning orders across alignment runs; (2) overly difficult examples significantly degrade performance across four LLMs and two datasets; and (3) the capacity of a model dictates its threshold for handling difficult examples, underscoring a critical relationship between data selection and model capacity. Building on this principle, we introduce Selective DPO, which filters out overly difficult examples. This simple adjustment improves alignment performance by 9-16% in win rates on the AlpacaEval 2 benchmark compared to the DPO baseline, suppressing a series of DPO variants with different algorithmic adjustments. Together, these results illuminate the importance of aligning data difficulty with model capacity, offering a transformative perspective for improving alignment strategies in LLMs. Code is available at https://github.com/glorgao/SelectiveDPO.

Yuxia Wang, Zenan Zhai, Haonan Li et al.

Many studies have demonstrated that large language models (LLMs) can produce harmful responses, exposing users to unexpected risks when LLMs are deployed. Previous studies have proposed comprehensive taxonomies of the risks posed by LLMs, as well as corresponding prompts that can be used to examine the safety mechanisms of LLMs. However, the focus has been almost exclusively on English, and little has been explored for other languages. Here we aim to bridge this gap. We first introduce a dataset for the safety evaluation of Chinese LLMs, and then extend it to two other scenarios that can be used to better identify false negative and false positive examples in terms of risky prompt rejections. We further present a set of fine-grained safety assessment criteria for each risk type, facilitating both manual annotation and automatic evaluation in terms of LLM response harmfulness. Our experiments on five LLMs show that region-specific risks are the prevalent type of risk, presenting the major issue with all Chinese LLMs we experimented with. Our data is available at https://github.com/Libr-AI/do-not-answer. Warning: this paper contains example data that may be offensive, harmful, or biased.

Jianshu She, Zhuohao Li, Zhemin Huang et al.

Chain-of-Thought (CoT) reasoning has demonstrated remarkable effectiveness in enhancing the reasoning abilities of large language models (LLMs). However, its efficiency remains a challenge due to the generation of excessive intermediate reasoning tokens, which introduce semantic redundancy and overly detailed reasoning steps. Moreover, computational expense and latency are significant concerns, as the cost scales with the number of output tokens, including those intermediate steps. In this work, we observe that most CoT tokens are unnecessary, and retaining only a small portion of them is sufficient for producing high-quality responses. Inspired by this, we propose HAWKEYE, a novel post-training and inference framework where a large model produces concise CoT instructions to guide a smaller model in response generation. HAWKEYE quantifies redundancy in CoT reasoning and distills high-density information via reinforcement learning. By leveraging these concise CoTs, HAWKEYE is able to expand responses while reducing token usage and computational cost significantly. Our evaluation shows that HAWKEYE can achieve comparable response quality using only 35% of the full CoTs, while improving clarity, coherence, and conciseness by approximately 10%. Furthermore, HAWKEYE can accelerate end-to-end reasoning by up to 3.4x on complex math tasks while reducing inference cost by up to 60%. HAWKEYE will be open-sourced and the models will be available soon.

Shengkun Tang, Liqun Ma, Haonan Li et al.

The typical Selective State-Space Model (SSM) used in Mamba addresses several limitations of Transformers, such as the quadratic computational complexity with respect to sequence length and the significant memory requirements during inference due to the key-value (KV) cache. However, the increasing size of Mamba models continues to pose challenges for training and deployment, particularly due to their substantial computational demands during both training and inference. In this work, we introduce $\texttt{Bi-Mamba}$, a scalable and powerful 1-bit Mamba architecture designed to enable more efficient large language models (LLMs), with model sizes of 780M, 1.3B, and 2.7B parameters. $\texttt{Bi-Mamba}$ models are trained from scratch on a standard LLM-scale dataset using an autoregressive distillation loss. Extensive experiments on language modeling benchmarks demonstrate that $\texttt{Bi-Mamba}$ achieves performance comparable to its full-precision (FP16 or BF16) counterparts, while outperforming post-training binarization (PTB) Mamba and binarization-aware training (BAT) Transformer baselines. Moreover, $\texttt{Bi-Mamba}$ drastically reduces memory usage and computational cost compared to the original Mamba. Our work pioneers a new line of linear-complexity LLMs under low-bit representation and provides the way for the design of specialized hardware optimized for efficient 1-bit Mamba-based models. Code and the pre-trained weights are available at https://github.com/Tangshengku/Bi-Mamba.

Zhengzhong Liu, Bowen Tan, Hongyi Wang et al.

We detail the training of the LLM360 K2-65B model, scaling up our 360-degree OPEN SOURCE approach to the largest and most powerful models under project LLM360. While open-source LLMs continue to advance, the answer to "How are the largest LLMs trained?" remains unclear within the community. The implementation details for such high-capacity models are often protected due to business considerations associated with their high cost. This lack of transparency prevents LLM researchers from leveraging valuable insights from prior experience, e.g., "What are the best practices for addressing loss spikes?" The LLM360 K2 project addresses this gap by providing full transparency and access to resources accumulated during the training of LLMs at the largest scale. This report highlights key elements of the K2 project, including our first model, K2 DIAMOND, a 65 billion-parameter LLM that surpasses LLaMA-65B and rivals LLaMA2-70B, while requiring fewer FLOPs and tokens. We detail the implementation steps and present a longitudinal analysis of K2 DIAMOND's capabilities throughout its training process. We also outline ongoing projects such as TXT360, setting the stage for future models in the series. By offering previously unavailable resources, the K2 project also resonates with the 360-degree OPEN SOURCE principles of transparency, reproducibility, and accessibility, which we believe are vital in the era of resource-intensive AI research.

Zheng Zhang, Haonan Li, Xingyu Li et al.

Bug bisection has been an important security task that aims to understand the range of software versions impacted by a bug, i.e., identifying the commit that introduced the bug. However, traditional patch-based bisection methods are faced with several significant barriers: For example, they assume that the bug-inducing commit (BIC) and the patch commit modify the same functions, which is not always true. They often rely solely on code changes, while the commit message frequently contains a wealth of vulnerability-related information. They are also based on simple heuristics (e.g., assuming the BIC initializes lines deleted in the patch) and lack any logical analysis of the vulnerability. In this paper, we make the observation that Large Language Models (LLMs) are well-positioned to break the barriers of existing solutions, e.g., comprehend both textual data and code in patches and commits. Unlike previous BIC identification approaches, which yield poor results, we propose a comprehensive multi-stage pipeline that leverages LLMs to: (1) fully utilize patch information, (2) compare multiple candidate commits in context, and (3) progressively narrow down the candidates through a series of down-selection steps. In our evaluation, we demonstrate that our approach achieves significantly better accuracy than the state-of-the-art solution by more than 38\%. Our results further confirm that the comprehensive multi-stage pipeline is essential, as it improves accuracy by 60\% over a baseline LLM-based bisection method.

Monojit Choudhury, Shivam Chauhan, Rocktim Jyoti Das et al.

Developing high-quality large language models (LLMs) for moderately resourced languages presents unique challenges in data availability, model adaptation, and evaluation. We introduce Llama-3-Nanda-10B-Chat, or Nanda for short, a state-of-the-art Hindi-centric instruction-tuned generative LLM, designed to push the boundaries of open-source Hindi language models. Built upon Llama-3-8B, Nanda incorporates continuous pre-training with expanded transformer blocks, leveraging the Llama Pro methodology. A key challenge was the limited availability of high-quality Hindi text data; we addressed this through rigorous data curation, augmentation, and strategic bilingual training, balancing Hindi and English corpora to optimize cross-linguistic knowledge transfer. With 10 billion parameters, Nanda stands among the top-performing open-source Hindi and multilingual models of similar scale, demonstrating significant advantages over many existing models. We provide an in-depth discussion of training strategies, fine-tuning techniques, safety alignment, and evaluation metrics, demonstrating how these approaches enabled Nanda to achieve state-of-the-art results. By open-sourcing Nanda, we aim to advance research in Hindi LLMs and support a wide range of real-world applications across academia, industry, and public services.

Zhoujun Cheng, Richard Fan, Shibo Hao et al.

K2-Think is a reasoning system that achieves state-of-the-art performance with a 32B parameter model, matching or surpassing much larger models like GPT-OSS 120B and DeepSeek v3.1. Built on the Qwen2.5 base model, our system shows that smaller models can compete at the highest levels by combining advanced post-training and test-time computation techniques. The approach is based on six key technical pillars: Long Chain-of-thought Supervised Finetuning, Reinforcement Learning with Verifiable Rewards (RLVR), Agentic planning prior to reasoning, Test-time Scaling, Speculative Decoding, and Inference-optimized Hardware, all using publicly available open-source datasets. K2-Think excels in mathematical reasoning, achieving state-of-the-art scores on public benchmarks for open-source models, while also performing strongly in other areas such as Code and Science. Our results confirm that a more parameter-efficient model like K2-Think 32B can compete with state-of-the-art systems through an integrated post-training recipe that includes long chain-of-thought training and strategic inference-time enhancements, making open-source reasoning systems more accessible and affordable. K2-Think is freely available at k2think.ai, offering best-in-class inference speeds of over 2,000 tokens per second per request via the Cerebras Wafer-Scale Engine.

Hanning Chen, Keyu Man, Kevin Zhu et al.

Identifying and addressing performance anti-patterns in machine learning (ML) models is critical for efficient training and inference, but it typically demands deep expertise spanning system infrastructure, ML models and kernel development. While large tech companies rely on dedicated ML infrastructure engineers to analyze torch traces and benchmarks, such resource-intensive workflows are largely inaccessible to computer vision researchers in general. Among the challenges, pinpointing problematic trace segments within lengthy execution traces remains the most time-consuming task, and is difficult to automate with current ML models, including LLMs. In this work, we present the first benchmark dataset specifically designed to evaluate and improve ML models' ability to detect anti patterns in traces. Our dataset contains over 600 PyTorch traces from diverse computer vision models classification, detection, segmentation, and generation collected across multiple hardware platforms. We also propose a novel iterative approach: a lightweight ML model first detects trace segments with anti patterns, followed by a large language model (LLM) for fine grained classification and targeted feedback. Experimental results demonstrate that our method significantly outperforms unsupervised clustering and rule based statistical techniques for detecting anti pattern regions. Our method also effectively compensates LLM's limited context length and reasoning inefficiencies.

K2 Team, Zhengzhong Liu, Liping Tang et al.

We introduce K2-V2, a 360-open LLM built from scratch as a superior base for reasoning adaptation, in addition to functions such as conversation and knowledge retrieval from general LLMs. It stands as the strongest fully open model, rivals open-weight leaders in its size class, outperforms Qwen2.5-72B and approaches the performance of Qwen3-235B. We actively infuse domain knowledge, reasoning, long-context, and tool use throughout the training process. This explicitly prepares the model for complex reasoning tasks. We demonstrate this potential using simple supervised fine-tuning, establishing a strong baseline that indicates significant headroom for advanced alignment. By releasing the full training history and data composition, we maximize the effectiveness of continuous training, a key open source production scenario. We release the model weights and signature LLM360 artifacts, such as complete training data, to empower the community with a capable, reasoning-centric foundation.

Jiahui Geng, Fengyu Cai, Shaobo Cui et al.

Code retrieval is essential in modern software development, as it boosts code reuse and accelerates debugging. However, current benchmarks primarily emphasize functional relevance while neglecting critical dimensions of software quality. Motivated by this gap, we introduce CoQuIR, the first large-scale, multilingual benchmark specifically designed to evaluate quality-aware code retrieval across four key dimensions: correctness, efficiency, security, and maintainability. CoQuIR provides fine-grained quality annotations for 42,725 queries and 134,907 code snippets in 11 programming languages, and is accompanied by two quality-centric evaluation metrics: Pairwise Preference Accuracy and Margin-based Ranking Score. Using CoQuIR, we benchmark 23 retrieval models, covering both open-source and proprietary systems, and find that even top-performing models frequently fail to distinguish buggy or insecure code from their more robust counterparts. Furthermore, we conduct preliminary investigations into training methods that explicitly encourage retrievers to recognize code quality. Using synthetic datasets, we demonstrate promising improvements in quality-aware metrics across various models, without sacrificing semantic relevance. Downstream code generation experiments further validate the effectiveness of our approach. Overall, our work highlights the importance of integrating quality signals into code retrieval systems, laying the groundwork for more trustworthy and robust software development tools.

Sukmin Yun, Haokun Lin, Rusiru Thushara et al.

Multimodal large language models (MLLMs) have shown impressive success across modalities such as image, video, and audio in a variety of understanding and generation tasks. However, current MLLMs are surprisingly poor at understanding webpage screenshots and generating their corresponding HTML code. To address this problem, we propose $\texttt{Web2Code}$, a benchmark consisting of a new large-scale webpage-to-code dataset for instruction tuning and an evaluation framework for the webpage understanding and HTML code translation abilities of MLLMs. For dataset construction, we leverage pretrained LLMs to enhance existing webpage-to-code datasets as well as generate a diverse pool of new webpages rendered into images. Specifically, the inputs are webpage images and instructions, while the responses are the webpage's HTML code. We further include diverse natural language QA pairs about the webpage content in the responses to enable a more comprehensive understanding of the web content. To evaluate model performance in these tasks, we develop an evaluation framework for testing MLLMs' abilities in webpage understanding and web-to-code generation. Extensive experiments show that our proposed dataset is beneficial not only to our proposed tasks but also in the general visual domain. We hope our work will contribute to the development of general MLLMs suitable for web-based content generation and task automation. Our data and code are available at https://github.com/MBZUAI-LLM/web2code.

Haonan Li, Martin Tomko, Timothy Baldwin

Answering real-world tourism questions that seek Point-of-Interest (POI) recommendations is challenging, as it requires both spatial and non-spatial reasoning, over a large candidate pool. The traditional method of encoding each pair of question and POI becomes inefficient when the number of candidates increases, making it infeasible for real-world applications. To overcome this, we propose treating the QA task as a dense vector retrieval problem, where we encode questions and POIs separately and retrieve the most relevant POIs for a question by utilizing embedding space similarity. We use pretrained language models (PLMs) to encode textual information, and train a location encoder to capture spatial information of POIs. Experiments on a real-world tourism QA dataset demonstrate that our approach is effective, efficient, and outperforms previous methods across all metrics. Enabled by the dense retrieval architecture, we further build a global evaluation baseline, expanding the search space by 20 times compared to previous work. We also explore several factors that impact on the model's performance through follow-up experiments. Our code and model are publicly available at https://github.com/haonan-li/LAMB.

Haonan Li, Fajri Koto, Minghao Wu et al.

Instruction tuning has shown great promise in improving the performance of large language models. However, research on multilingual instruction tuning has been limited due to the scarcity of high-quality instruction-response datasets across different languages. To bridge this gap, we present Bactrian-X, a comprehensive multilingual parallel dataset of 3.4 million instruction-response pairs across 52 languages. Leveraging this dataset, we train a set of adapters using low-rank adaptation (LoRA), which are lightweight components that seamlessly integrate with large language models. These adapters have a substantially lower parameter count than the base model, making them easily replaceable and usable as plug-ins for different languages or language groups. Extensive experiments in various multilingual evaluation settings demonstrate that models derived from LoRA-based training over Bactrian-X outperform both the vanilla models and existing instruction-tuned models. The code and models are publicly available at https://github.com/mbzuai-nlp/bactrian-x

Xiangru Lian, Binhang Yuan, Xuefeng Zhu et al.

Deep learning based models have dominated the current landscape of production recommender systems. Furthermore, recent years have witnessed an exponential growth of the model scale--from Google's 2016 model with 1 billion parameters to the latest Facebook's model with 12 trillion parameters. Significant quality boost has come with each jump of the model capacity, which makes us believe the era of 100 trillion parameters is around the corner. However, the training of such models is challenging even within industrial scale data centers. This difficulty is inherited from the staggering heterogeneity of the training computation--the model's embedding layer could include more than 99.99% of the total model size, which is extremely memory-intensive; while the rest neural network is increasingly computation-intensive. To support the training of such huge models, an efficient distributed training system is in urgent need. In this paper, we resolve this challenge by careful co-design of both the optimization algorithm and the distributed system architecture. Specifically, in order to ensure both the training efficiency and the training accuracy, we design a novel hybrid training algorithm, where the embedding layer and the dense neural network are handled by different synchronization mechanisms; then we build a system called Persia (short for parallel recommendation training system with hybrid acceleration) to support this hybrid training algorithm. Both theoretical demonstration and empirical study up to 100 trillion parameters have conducted to justified the system design and implementation of Persia. We make Persia publicly available (at https://github.com/PersiaML/Persia) so that anyone would be able to easily train a recommender model at the scale of 100 trillion parameters.

Ekaterina Fadeeva, Aleksandr Rubashevskii, Artem Shelmanov et al.

Large language models (LLMs) are notorious for hallucinating, i.e., producing erroneous claims in their output. Such hallucinations can be dangerous, as occasional factual inaccuracies in the generated text might be obscured by the rest of the output being generally factually correct, making it extremely hard for the users to spot them. Current services that leverage LLMs usually do not provide any means for detecting unreliable generations. Here, we aim to bridge this gap. In particular, we propose a novel fact-checking and hallucination detection pipeline based on token-level uncertainty quantification. Uncertainty scores leverage information encapsulated in the output of a neural network or its layers to detect unreliable predictions, and we show that they can be used to fact-check the atomic claims in the LLM output. Moreover, we present a novel token-level uncertainty quantification method that removes the impact of uncertainty about what claim to generate on the current step and what surface form to use. Our method Claim Conditioned Probability (CCP) measures only the uncertainty of a particular claim value expressed by the model. Experiments on the task of biography generation demonstrate strong improvements for CCP compared to the baselines for seven LLMs and four languages. Human evaluation reveals that the fact-checking pipeline based on uncertainty quantification is competitive with a fact-checking tool that leverages external knowledge.

Lizhi Lin, Honglin Mu, Zenan Zhai et al.

Generative models are rapidly gaining popularity and being integrated into everyday applications, raising concerns over their safe use as various vulnerabilities are exposed. In light of this, the field of red teaming is undergoing fast-paced growth, highlighting the need for a comprehensive survey covering the entire pipeline and addressing emerging topics. Our extensive survey, which examines over 120 papers, introduces a taxonomy of fine-grained attack strategies grounded in the inherent capabilities of language models. Additionally, we have developed the "searcher" framework to unify various automatic red teaming approaches. Moreover, our survey covers novel areas including multimodal attacks and defenses, risks around LLM-based agents, overkill of harmless queries, and the balance between harmlessness and helpfulness.

Rocktim Jyoti Das, Simeon Emilov Hristov, Haonan Li et al.

We introduce EXAMS-V, a new challenging multi-discipline multimodal multilingual exam benchmark for evaluating vision language models. It consists of 20,932 multiple-choice questions across 20 school disciplines covering natural science, social science, and other miscellaneous studies, e.g., religion, fine arts, business, etc. EXAMS-V includes a variety of multimodal features such as text, images, tables, figures, diagrams, maps, scientific symbols, and equations. The questions come in 11 languages from 7 language families. Unlike existing benchmarks, EXAMS-V is uniquely curated by gathering school exam questions from various countries, with a variety of education systems. This distinctive approach calls for intricate reasoning across diverse languages and relies on region-specific knowledge. Solving the problems in the dataset requires advanced perception and joint reasoning over the text and the visual content of the image. Our evaluation results demonstrate that this is a challenging dataset, which is difficult even for advanced vision-text models such as GPT-4V and Gemini; this underscores the inherent complexity of the dataset and its significance as a future benchmark.

Renxi Wang, Haonan Li, Xudong Han et al.

Large language models (LLMs) have achieved success in acting as agents, which interact with environments through tools such as search engines. However, LLMs are optimized for language generation instead of tool use during training or alignment, limiting their effectiveness as agents. To resolve this problem, previous work has first collected interaction trajectories between LLMs and environments, using only trajectories that successfully finished the task to fine-tune smaller models, making fine-tuning data scarce and acquiring it both difficult and costly. Discarding failed trajectories also leads to significant wastage of data and resources and limits the possible optimization paths during fine-tuning. In this paper, we argue that unsuccessful trajectories offer valuable insights, and LLMs can learn from these trajectories through appropriate quality control and fine-tuning strategies. By simply adding a prefix or suffix that tells the model whether to generate a successful trajectory during training, we improve model performance by a large margin on mathematical reasoning, multi-hop question answering, and strategic question answering tasks. We further analyze the inference results and find that our method provides a better trade-off between valuable information and errors in unsuccessful trajectories. To our knowledge, we are the first to demonstrate the value of negative trajectories and their application in agent-tunning scenarios. Our findings offer guidance for developing better agent-tuning methods and low-resource data usage techniques.

Haonan Li, Tianjun Sun, Yongqing Wang et al.

Multi-server MCP agents create an information-flow control problem: faithful tool composition can turn individually benign read/write permissions into cross-boundary credential propagation -- a structural side effect of workflow topology, not necessarily malicious model behavior. We present MCPHunt, to our knowledge the first controlled benchmark that isolates non-adversarial, verbatim credential propagation across multi-server MCP trust boundaries, with three methodological contributions: (1) canary-based taint tracking that reduces propagation detection to objective string matching; (2) an environment-controlled coverage design with risky, benign, and hard-negative conditions that validates pipeline soundness and controls for credential-format confounds; (3) CRS stratification that disentangles task-mandated propagation (faithful execution of verbatim-transfer instructions) from policy-violating propagation (credentials included despite the option to redact). Across 3,615 main-benchmark traces from 5 models spanning 147 tasks and 9 mechanism families, policy-violating propagation rates reach 11.5--41.3% across all models. This propagation is pathway-specific (25x cross-mechanism range) and concentrated in browser-mediated data flows; hard-negative controls provide evidence that production-format credentials are not necessary -- prompt-directed cross-boundary data flow is sufficient. A prompt-mitigation study across 3 models reduces policy-violating propagation by up to 97% while preserving 80.5% utility, but effectiveness varies with instruction-following capability -- suggesting that prompt-level defenses alone may not suffice. Code, traces, and labeling pipeline are released under MIT and CC BY 4.0.

Yilin Geng, Haonan Li, Honglin Mu et al.

Large language models (LLMs) are increasingly deployed with hierarchical instruction schemes, where certain instructions (e.g., system-level directives) are expected to take precedence over others (e.g., user messages). Yet, we lack a systematic understanding of how effectively these hierarchical control mechanisms work. We introduce a systematic evaluation framework based on constraint prioritization to assess how well LLMs enforce instruction hierarchies. Our experiments across six state-of-the-art LLMs reveal that models struggle with consistent instruction prioritization, even for simple formatting conflicts. We find that the widely-adopted system/user prompt separation fails to establish a reliable instruction hierarchy, and models exhibit strong inherent biases toward certain constraint types regardless of their priority designation. We find that LLMs more reliably obey constraints framed through natural social hierarchies (e.g., authority, expertise, consensus) than system/user roles, which suggests that pretraining-derived social structures act as latent control priors, with potentially stronger influence than post-training guardrails.

Zhuohan Xie, Daniil Orel, Rushil Thareja et al.

Multi-step symbolic reasoning is essential for robust financial analysis; yet, current benchmarks largely overlook this capability. Existing datasets such as FinQA and ConvFinQA emphasize final numerical answers while neglecting the intermediate reasoning required for transparency and verification. To address this gap, we introduce FinChain, the first benchmark specifically designed for verifiable Chain-of-Thought (CoT) evaluation in finance. FinChain spans 58 topics across 12 financial domains, each represented by parameterized symbolic templates with executable Python traces that enable fully machine-verifiable reasoning and scalable, contamination-free data generation. To assess reasoning capacity, we propose ChainEval, a dynamic alignment metric that jointly evaluates both the final-answer correctness and the step-level reasoning consistency. Evaluating 26 leading LLMs reveals that even frontier proprietary systems exhibit clear limitations in symbolic financial reasoning, while domain-adapted and math-enhanced fine-tuned models substantially narrow this gap. Overall, FinChain exposes persistent weaknesses in multi-step financial reasoning and provides a foundation for developing trustworthy, interpretable, and verifiable financial AI.

Haonan Li, Xudong Han, Zenan Zhai et al.

To address this gap, we introduce Libra-Leaderboard, a comprehensive framework designed to rank LLMs through a balanced evaluation of performance and safety. Combining a dynamic leaderboard with an interactive LLM arena, Libra-Leaderboard encourages the joint optimization of capability and safety. Unlike traditional approaches that average performance and safety metrics, Libra-Leaderboard uses a distance-to-optimal-score method to calculate the overall rankings. This approach incentivizes models to achieve a balance rather than excelling in one dimension at the expense of some other ones. In the first release, Libra-Leaderboard evaluates 26 mainstream LLMs from 14 leading organizations, identifying critical safety challenges even in state-of-the-art models.

Zenan Zhai, Hao Li, Xudong Han et al.

Recent advances in large language models (LLMs) have shown that they can answer questions requiring complex reasoning. However, their ability to identify and respond to text containing logical fallacies or deliberately misleading premises remains less studied. To address this gap, we introduce RuozhiBench, a bilingual dataset comprising 677 carefully curated questions that contain various forms of deceptive reasoning, meticulously crafted through extensive human effort and expert review. In a comprehensive evaluation of 17 LLMs from 5 Series over RuozhiBench using both open-ended and two-choice formats, we conduct extensive analyses on evaluation protocols and result patterns. Despite their high scores on conventional benchmarks, these models showed limited ability to detect and reason correctly about logical fallacies, with even the best-performing model, Claude-3-haiku, achieving only 62% accuracy compared to the human of more than 90%.

Xingjun Ma, Yifeng Gao, Yixu Wang et al.

The rapid advancement of large models, driven by their exceptional abilities in learning and generalization through large-scale pre-training, has reshaped the landscape of Artificial Intelligence (AI). These models are now foundational to a wide range of applications, including conversational AI, recommendation systems, autonomous driving, content generation, medical diagnostics, and scientific discovery. However, their widespread deployment also exposes them to significant safety risks, raising concerns about robustness, reliability, and ethical implications. This survey provides a systematic review of current safety research on large models, covering Vision Foundation Models (VFMs), Large Language Models (LLMs), Vision-Language Pre-training (VLP) models, Vision-Language Models (VLMs), Diffusion Models (DMs), and large-model-powered Agents. Our contributions are summarized as follows: (1) We present a comprehensive taxonomy of safety threats to these models, including adversarial attacks, data poisoning, backdoor attacks, jailbreak and prompt injection attacks, energy-latency attacks, data and model extraction attacks, and emerging agent-specific threats. (2) We review defense strategies proposed for each type of attacks if available and summarize the commonly used datasets and benchmarks for safety research. (3) Building on this, we identify and discuss the open challenges in large model safety, emphasizing the need for comprehensive safety evaluations, scalable and effective defense mechanisms, and sustainable data practices. More importantly, we highlight the necessity of collective efforts from the research community and international collaboration. Our work can serve as a useful reference for researchers and practitioners, fostering the ongoing development of comprehensive defense systems and platforms to safeguard AI models.

Renxi Wang, Haonan Li, Minghao Wu et al.

Instruction tuning significantly enhances the performance of large language models (LLMs) across various tasks. However, the procedure to optimizing the mixing of instruction datasets for LLM fine-tuning is still poorly understood. This study categorizes instructions into three primary types: NLP downstream tasks, coding, and general chat. We explore the effects of instruction tuning on different combinations of datasets on LLM performance, and find that certain instruction types are more advantageous for specific applications but can negatively impact other areas. This work provides insights into instruction mixtures, laying the foundations for future research.

Shuo Jiang, Haonan Li, Ruochen Ren et al.

Cutting-edge robot learning techniques including foundation models and imitation learning from humans all pose huge demands on large-scale and high-quality datasets which constitute one of the bottleneck in the general intelligent robot fields. This paper presents the Kaiwu multimodal dataset to address the missing real-world synchronized multimodal data problems in the sophisticated assembling scenario,especially with dynamics information and its fine-grained labelling. The dataset first provides an integration of human,environment and robot data collection framework with 20 subjects and 30 interaction objects resulting in totally 11,664 instances of integrated actions. For each of the demonstration,hand motions,operation pressures,sounds of the assembling process,multi-view videos, high-precision motion capture information,eye gaze with first-person videos,electromyography signals are all recorded. Fine-grained multi-level annotation based on absolute timestamp,and semantic segmentation labelling are performed. Kaiwu dataset aims to facilitate robot learning,dexterous manipulation,human intention investigation and human-robot collaboration research.

Rawan Al-Matham, Kareem Darwish, Raghad Al-Rasheed et al.

The impressive advancement of Large Language Models (LLMs) in English has not been matched across all languages. In particular, LLM performance in Arabic lags behind, due to data scarcity, linguistic diversity of Arabic and its dialects, morphological complexity, etc. Progress is further hindered by the quality of Arabic benchmarks, which typically rely on static, publicly available data, lack comprehensive task coverage, or do not provide dedicated platforms with blind test sets. This makes it challenging to measure actual progress and to mitigate data contamination. Here, we aim to bridge these gaps. In particular, we introduce BALSAM, a comprehensive, community-driven benchmark aimed at advancing Arabic LLM development and evaluation. It includes 78 NLP tasks from 14 broad categories, with 52K examples divided into 37K test and 15K development, and a centralized, transparent platform for blind evaluation. We envision BALSAM as a unifying platform that sets standards and promotes collaborative research to advance Arabic LLM capabilities.

Renxi Wang, Rifo Ahmad Genadi, Bilal El Bouardi et al.

Language model (LM) agents have gained significant attention for their ability to autonomously complete tasks through interactions with environments, tools, and APIs. LM agents are primarily built with prompt engineering or supervised finetuning. At the same time, reinforcement learning (RL) has been explored to enhance LM's capabilities, such as reasoning and factuality. However, the combination of the LM agents and reinforcement learning (Agent-RL) remains underexplored and lacks systematic study. To this end, we built AgentFly, a scalable and extensible Agent-RL framework designed to empower LM agents with a variety of RL algorithms. Our framework supports multi-turn interactions by adapting traditional RL methods with token-level masking. It features a decorator-based interface for defining tools and reward functions, enabling seamless extension and ease of use. To support high-throughput training, we implement asynchronous execution of tool calls and reward computations, and design a centralized resource management system for scalable environment coordination. We also provide a suite of prebuilt tools and environments, demonstrating the framework's effectiveness through successful agent training across multiple tasks.

Haonan Li, Patrick P. K. Chen, Yitong Zhou

With the rapid advancement of technologies such as virtual reality, augmented reality, and gesture control, users expect interactions with computer interfaces to be more natural and intuitive. Existing visual algorithms often struggle to accomplish advanced human-computer interaction tasks, necessitating accurate and reliable absolute spatial prediction methods. Moreover, dealing with complex scenes and occlusions in monocular images poses entirely new challenges. This study proposes a network model that performs parallel processing of root-relative grids and root recovery tasks. The model enables the recovery of 3D hand meshes in camera space from monocular RGB images. To facilitate end-to-end training, we utilize an implicit learning approach for 2D heatmaps, enhancing the compatibility of 2D cues across different subtasks. Incorporate the Inception concept into spectral graph convolutional network to explore relative mesh of root, and integrate it with the locally detailed and globally attentive method designed for root recovery exploration. This approach improves the model's predictive performance in complex environments and self-occluded scenes. Through evaluation on the large-scale hand dataset FreiHAND, we have demonstrated that our proposed model is comparable with state-of-the-art models. This study contributes to the advancement of techniques for accurate and reliable absolute spatial prediction in various human-computer interaction applications.

Haonan Li, Hang Zhang, Kexin Pei et al.

Static analysis plays a crucial role in software vulnerability detection, yet faces a persistent precision-scalability tradeoff. In large codebases like the Linux kernel, traditional static analysis tools often generate excessive false positives due to simplified vulnerability modeling and overapproximation of path and data constraints. While large language models (LLMs) demonstrate promising code understanding capabilities, their direct application to program analysis remains unreliable due to inherent reasoning limitations. We introduce BugLens, a post-refinement framework that significantly enhances static analysis precision for bug detection. BugLens guides LLMs through structured reasoning steps to assess security impact and validate constraints from the source code. When evaluated on Linux kernel taint-style bugs detected by static analysis tools, BugLens improves precision approximately 7-fold (from 0.10 to 0.72), substantially reducing false positives while uncovering four previously unreported vulnerabilities. Our results demonstrate that a well-structured, fully automated LLM-based workflow can effectively complement and enhance traditional static analysis techniques.

Qiyuan Xu, Xiaokun Luan, Renxi Wang et al.

Theorem proving is fundamental to program verification, where the automated proof of Verification Conditions (VCs) remains a primary bottleneck. Real-world program verification frequently encounters hard VCs that existing Automated Theorem Provers (ATPs) cannot prove, leading to a critical need for extensive manual proofs that burden practical application. While Neural Theorem Proving (NTP) has achieved significant success in mathematical competitions, demonstrating the potential of machine learning approaches to formal reasoning, its application to program verification--particularly VC proving--remains largely unexplored. Despite existing work on annotation synthesis and verification-related theorem proving, no benchmark has specifically targeted this fundamental bottleneck: automated VC proving. This work introduces Neural Theorem Proving for Verification Conditions (NTP4VC), presenting the first real-world multi-language benchmark for this task. From real-world projects such as Linux and Contiki-OS kernel, our benchmark leverages industrial pipelines (Why3 and Frama-C) to generate semantically equivalent test cases across formal languages of Isabelle, Lean, and Rocq. We evaluate large language models (LLMs), both general-purpose and those fine-tuned for theorem proving, on NTP4VC. Results indicate that although LLMs show promise in VC proving, significant challenges remain for program verification, highlighting a large gap and opportunity for future research.

Chengqian Gao, Haonan Li, Taylor W. Killian et al.

Concise reasoning in large language models seeks to generate only essential intermediate steps needed to arrive at a final answer, thereby alleviating issues of overthinking. Most proposed approaches hinge on carefully hand-crafted heuristics, struggling to balance concision with performance, often failing to adapt across domains and model scales. In this work, we address these challenges by introducing a principled and pragmatic strategy, performance-aware length updating (PALU). As a principled algorithm, PALU formulates concise reasoning as a constrained optimization problem, minimizing response length subject to a performance constraint, and then applies Lagrangian optimization to convert it into a tractable unconstrained problem. As a pragmatic solution, PALU streamlines complicated update rules through three approximations: (i) estimating performance with off-policy rollouts, (ii) truncating the Lagrange multiplier to two extremes, and (iii) replacing gradient-based updates with quantile-driven length adjustments. PALU reduces output length by 65% while improving accuracy by 15% when applied to DeepSeek-Distill-Qwen-1.5B, averaged over five benchmarks, outperforming a range of alternative methods. Furthermore, PALU is demonstrated to adapt across both domain (logic, STEM and math) and model scale (1.5B, 7B, 14B) entrenching the algorithm as a practical and effective concise reasoning approach.

Yang Li, Chenyu Wang, Tingrui Wang et al.

Black-box adversarial attacks remain challenging due to limited access to model internals. Existing methods often depend on specific network architectures or require numerous queries, resulting in limited cross-architecture transferability and high query costs. To address these limitations, we propose JAD, a latent diffusion model framework for black-box adversarial attacks. JAD generates adversarial examples by leveraging a latent diffusion model guided by attention maps distilled from both a convolutional neural network (CNN) and a Vision Transformer (ViT) models. By focusing on image regions that are commonly sensitive across architectures, this approach crafts adversarial perturbations that transfer effectively between different model types. This joint attention distillation strategy enables JAD to be architecture-agnostic, achieving superior attack generalization across diverse models. Moreover, the generative nature of the diffusion framework yields high adversarial sample generation efficiency by reducing reliance on iterative queries. Experiments demonstrate that JAD offers improved attack generalization, generation efficiency, and cross-architecture transferability compared to existing methods, providing a promising and effective paradigm for black-box adversarial attacks.

Renxi Wang, Honglin Mu, Liqun Ma et al.

Evaluating large language models' (LLMs) long-context understanding capabilities remains challenging. We present SCALAR (Scientific Citation-based Live Assessment of Long-context Academic Reasoning), a novel benchmark that leverages academic papers and their citation networks. SCALAR features automatic generation of high-quality ground truth labels without human annotation, controllable difficulty levels, and a dynamic updating mechanism that prevents data contamination. Using ICLR 2025 papers, we evaluate 8 state-of-the-art LLMs, revealing key insights about their capabilities and limitations in processing long scientific documents across different context lengths and reasoning types. Our benchmark provides a reliable and sustainable way to track progress in long-context understanding as LLM capabilities evolve.