Daily Papers

To be informative, an evaluation must measure how well systems generalize to realistic unseen data. We identify limitations of and propose improvements to current evaluations of text-to-SQL systems. First, we compare human-generated and automatically generated questions, characterizing properties of queries necessary for real-world applications. To facilitate evaluation on multiple datasets, we release standardized and improved versions of seven existing datasets and one new text-to-SQL dataset. Second, we show that the current division of data into training and test sets measures robustness to variations in the way questions are asked, but only partially tests how well systems generalize to new queries; therefore, we propose a complementary dataset split for evaluation of future work. Finally, we demonstrate how the common practice of anonymizing variables during evaluation removes an important challenge of the task. Our observations highlight key difficulties, and our methodology enables effective measurement of future development.

Objective. Clinical AI documentation systems require evaluation methodologies that are clinically valid, economically viable, and sensitive to iterative changes. Methods requiring expert review per scoring instance are too slow and expensive for safe, iterative deployment. We present a case-specific, clinician-authored rubric methodology for clinical AI evaluation and examine whether LLM-generated rubrics can approximate clinician agreement. Materials and Methods. Twenty clinicians authored 1,646 rubrics for 823 clinical cases (736 real-world, 87 synthetic) across primary care, psychiatry, oncology, and behavioral health. Each rubric was validated by confirming that an LLM-based scoring agent consistently scored clinician-preferred outputs higher than rejected ones. Seven versions of an EHR-embedded AI agent for clinicians were evaluated across all cases. Results. Clinician-authored rubrics discriminated effectively between high- and low-quality outputs (median score gap: 82.9%) with high scoring stability (median range: 0.00%). Median scores improved from 84% to 95%. In later experiments, clinician-LLM ranking agreement (tau: 0.42-0.46) matched or exceeded clinician-clinician agreement (tau: 0.38-0.43), attributable to both ceiling compression and LLM rubric improvement. Discussion. This convergence supports incorporating LLM rubrics alongside clinician-authored ones. At roughly 1,000 times lower cost, LLM rubrics enable substantially greater evaluation coverage, while continued clinical authorship grounds evaluation in expert judgment. Ceiling compression poses a methodological challenge for future inter-rater agreement studies. Conclusion. Case-specific rubrics offer a path for clinical AI evaluation that preserves expert judgment while enabling automation at three orders lower cost. Clinician-authored rubrics establish the baseline against which LLM rubrics are validated.

Speech synthesis has come a long way as current text-to-speech (TTS) models can now generate natural human-sounding speech. However, most of the TTS research focuses on using adult speech data and there has been very limited work done on child speech synthesis. This study developed and validated a training pipeline for fine-tuning state-of-the-art (SOTA) neural TTS models using child speech datasets. This approach adopts a multi-speaker TTS retuning workflow to provide a transfer-learning pipeline. A publicly available child speech dataset was cleaned to provide a smaller subset of approximately 19 hours, which formed the basis of our fine-tuning experiments. Both subjective and objective evaluations were performed using a pretrained MOSNet for objective evaluation and a novel subjective framework for mean opinion score (MOS) evaluations. Subjective evaluations achieved the MOS of 3.95 for speech intelligibility, 3.89 for voice naturalness, and 3.96 for voice consistency. Objective evaluation using a pretrained MOSNet showed a strong correlation between real and synthetic child voices. Speaker similarity was also verified by calculating the cosine similarity between the embeddings of utterances. An automatic speech recognition (ASR) model is also used to provide a word error rate (WER) comparison between the real and synthetic child voices. The final trained TTS model was able to synthesize child-like speech from reference audio samples as short as 5 seconds.

Over the past few years, one of the most notable advancements in AI research has been in foundation models (FMs), headlined by the rise of language models (LMs). As the models' size increases, LMs demonstrate enhancements in measurable aspects and the development of new qualitative features. However, despite researchers' attention and the rapid growth in LM application, the capabilities, limitations, and associated risks still need to be better understood. To address these issues, we introduce an open Multimodal Evaluation of Russian-language Architectures (MERA), a new instruction benchmark for evaluating foundation models oriented towards the Russian language. The benchmark encompasses 21 evaluation tasks for generative models in 11 skill domains and is designed as a black-box test to ensure the exclusion of data leakage. The paper introduces a methodology to evaluate FMs and LMs in zero- and few-shot fixed instruction settings that can be extended to other modalities. We propose an evaluation methodology, an open-source code base for the MERA assessment, and a leaderboard with a submission system. We evaluate open LMs as baselines and find that they are still far behind the human level. We publicly release MERA to guide forthcoming research, anticipate groundbreaking model features, standardize the evaluation procedure, and address potential societal drawbacks.

Patent novelty search systems are critical to IP protection and innovation assessment; their retrieval accuracy directly impacts patent quality. We propose a comprehensive evaluation methodology that builds high-quality, reproducible datasets from examiner citations and X-type citations extracted from technically consistent family patents, and evaluates systems using invention descriptions as inputs. Using Top-k Detection Rate and Recall as core metrics, we further conduct multi-dimensional analyses by language, technical field (IPC), and filing jurisdiction. Experiments show the method effectively exposes performance differences across scenarios and offers actionable evidence for system improvement. The framework is scalable and practical, providing a useful reference for development and optimization of patent novelty search systems

Natural language inference (NLI) is the task of determining if a natural language hypothesis can be inferred from a given premise in a justifiable manner. NLI was proposed as a benchmark task for natural language understanding. Existing models perform well at standard datasets for NLI, achieving impressive results across different genres of text. However, the extent to which these models understand the semantic content of sentences is unclear. In this work, we propose an evaluation methodology consisting of automatically constructed "stress tests" that allow us to examine whether systems have the ability to make real inferential decisions. Our evaluation of six sentence-encoder models on these stress tests reveals strengths and weaknesses of these models with respect to challenging linguistic phenomena, and suggests important directions for future work in this area.

RTK-SLAM systems integrate simultaneous localization and mapping (SLAM) with real-time kinematic (RTK) GNSS positioning, promising both relative consistency and globally referenced coordinates for efficient georeferenced surveying. A critical and underappreciated issue is that the standard evaluation metric, Absolute Trajectory Error (ATE), first fits an optimal rigid-body transformation between the estimated trajectory and reference before computing errors. This so-called SE(3) alignment absorbs global drift and systematic errors, making trajectories appear more accurate than they are in practice, and is unsuitable for evaluating the global accuracy of RTK-SLAM. We present a geodetically referenced dataset and evaluation methodology that expose this gap. A key design principle is that the RTK receiver is used solely as a system input, while ground truth is established independently via a geodetic total station. This separation is absent from all existing datasets, where GNSS typically serves as (part of) the ground truth. The dataset is collected with a handheld RTK-SLAM device, comprising two scenes. We evaluate LiDAR-inertial, visual-inertial, and LiDAR-visual-inertial RTK-SLAM systems alongside standalone RTK, reporting direct global accuracy and SE(3)-aligned relative accuracy to make the gap explicit. Results show that SE(3) alignment can underestimate absolute positioning error by up to 76\%. RTK-SLAM achieves centimeter-level absolute accuracy in open-sky conditions and maintains decimeter-level global accuracy indoors, where standalone RTK degrades to tens of meters. The dataset, calibration files, and evaluation scripts are publicly available at https://rtk-slam-dataset.github.io/.

We introduce POLLUX, a comprehensive open-source benchmark designed to evaluate the generative capabilities of large language models (LLMs) in Russian. Our main contribution is a novel evaluation methodology that enhances the interpretability of LLM assessment. For each task type, we define a set of detailed criteria and develop a scoring protocol where models evaluate responses and provide justifications for their ratings. This enables transparent, criteria-driven evaluation beyond traditional resource-consuming, side-by-side human comparisons. POLLUX includes a detailed, fine-grained taxonomy of 35 task types covering diverse generative domains such as code generation, creative writing, and practical assistant use cases, totaling 2,100 manually crafted and professionally authored prompts. Each task is categorized by difficulty (easy/medium/hard), with experts constructing the dataset entirely from scratch. We also release a family of LLM-as-a-Judge (7B and 32B) evaluators trained for nuanced assessment of generative outputs. This approach provides scalable, interpretable evaluation and annotation tools for model development, effectively replacing costly and less precise human judgments.

We present ATANT (Automated Test for Acceptance of Narrative Truth), an open evaluation framework for measuring continuity in AI systems: the ability to persist, update, disambiguate, and reconstruct meaningful context across time. While the AI industry has produced memory components (RAG pipelines, vector databases, long context windows, profile layers), no published framework formally defines or measures whether these components produce genuine continuity. We define continuity as a system property with 7 required properties, introduce a 10-checkpoint evaluation methodology that operates without an LLM in the evaluation loop, and present a narrative test corpus of 250 stories comprising 1,835 verification questions across 6 life domains. We evaluate a reference implementation across 5 test suite iterations, progressing from 58% (legacy architecture) to 100% in isolated mode (250 stories) and 100% in 50-story cumulative mode, with 96% at 250-story cumulative scale. The cumulative result is the primary measure: when 250 distinct life narratives coexist in the same database, the system must retrieve the correct fact for the correct context without cross-contamination. ATANT is system-agnostic, model-independent, and designed as a sequenced methodology for building and validating continuity systems. The framework specification, example stories, and evaluation protocol are available at https://github.com/Kenotic-Labs/ATANT. The full 250-story corpus will be released incrementally.

Battles, or side-by-side comparisons in so called arenas that elicit human preferences, have emerged as a popular approach to assessing the output quality of LLMs. Recently, this idea has been extended to retrieval-augmented generation (RAG) systems. While undoubtedly representing an advance in evaluation, battles have at least two drawbacks, particularly in the context of complex information-seeking queries: they are neither explanatory nor diagnostic. Recently, the nugget evaluation methodology has emerged as a promising approach to evaluate the quality of RAG answers. Nuggets decompose long-form LLM-generated answers into atomic facts, highlighting important pieces of information necessary in a "good" response. In this work, we apply our AutoNuggetizer framework to analyze data from roughly 7K Search Arena battles provided by LMArena in a fully automatic manner. Our results show a significant correlation between nugget scores and human preferences, showcasing promise in our approach to explainable and diagnostic system evaluations.

In this article we introduce the Arcade Learning Environment (ALE): both a challenge problem and a platform and methodology for evaluating the development of general, domain-independent AI technology. ALE provides an interface to hundreds of Atari 2600 game environments, each one different, interesting, and designed to be a challenge for human players. ALE presents significant research challenges for reinforcement learning, model learning, model-based planning, imitation learning, transfer learning, and intrinsic motivation. Most importantly, it provides a rigorous testbed for evaluating and comparing approaches to these problems. We illustrate the promise of ALE by developing and benchmarking domain-independent agents designed using well-established AI techniques for both reinforcement learning and planning. In doing so, we also propose an evaluation methodology made possible by ALE, reporting empirical results on over 55 different games. All of the software, including the benchmark agents, is publicly available.

If you had an AI Whale-to-English translator, how could you validate whether or not it is working? Does one need to interact with the animals or rely on grounded observations such as temperature? We provide theoretical and proof-of-concept experimental evidence suggesting that interaction and even observations may not be necessary for sufficiently complex languages. One may be able to evaluate translators solely by their English outputs, offering potential advantages in terms of safety, ethics, and cost. This is an instance of machine translation quality evaluation (MTQE) without any reference translations available. A key challenge is identifying ``hallucinations,'' false translations which may appear fluent and plausible. We propose using segment-by-segment translation together with the classic NLP shuffle test to evaluate translators. The idea is to translate animal communication, turn by turn, and evaluate how often the resulting translations make more sense in order than permuted. Proof-of-concept experiments on data-scarce human languages and constructed languages demonstrate the potential utility of this evaluation methodology. These human-language experiments serve solely to validate our reference-free metric under data scarcity. It is found to correlate highly with a standard evaluation based on reference translations, which are available in our experiments. We also perform a theoretical analysis suggesting that interaction may not be necessary nor efficient in the early stages of learning to translate.

ATANT v1.0 (arXiv:2604.06710) defined continuity as a system property with 7 required properties and introduced a 10-checkpoint, LLM-free evaluation methodology validated on a 250-story corpus. Since publication, a recurring reviewer and practitioner question has concerned not the framework itself but its relationship to a wider set of memory evaluations: LOCOMO, LongMemEval, BEAM, MemoryBench, Zep's evaluation suite, Letta/MemGPT's evaluations, and RULER. This companion paper, v1.1, does not modify the v1.0 standard. It closes a related-work gap that v1.0 left brief under page limits. We show by structural analysis that none of these benchmarks measures continuity as defined in v1.0: of the 7 required properties, the median existing eval covers 1 property, the mean covers 0.43 when partial credit is scored at 0.5, and no eval covers more than 2. We provide a cell-by-cell property-coverage matrix, identify methodological defects specific to each benchmark (including an empty-gold scoring bug in the LOCOMO reference implementation that renders 23% of its corpus unscorable by construction), and publish our reference implementation's LOCOMO score (8.8%) alongside the structural reason that number is uninformative about continuity. We publish our 8.8% LOCOMO score alongside our 96% ATANT cumulative-scale score as a calibration pair: the 87-point divergence is evidence that the two benchmarks measure different properties, not that one system is an order of magnitude better than another. The position v1.1 takes is not adversarial: each benchmark measures a real capability. The claim is that none of them can adjudicate continuity, and conflating them with continuity evaluation has led the field to under-invest in the properties v1.0 names.

Large language models (LLMs) have shown impressive capabilities across various natural language tasks. However, evaluating their alignment with human preferences remains a challenge. To this end, we propose a comprehensive human evaluation framework to assess LLMs' proficiency in following instructions on diverse real-world tasks. We construct a hierarchical task tree encompassing 7 major areas covering over 200 categories and over 800 tasks, which covers diverse capabilities such as question answering, reasoning, multiturn dialogue, and text generation, to evaluate LLMs in a comprehensive and in-depth manner. We also design detailed evaluation standards and processes to facilitate consistent, unbiased judgments from human evaluators. A test set of over 3,000 instances is released, spanning different difficulty levels and knowledge domains. Our work provides a standardized methodology to evaluate human alignment in LLMs for both English and Chinese. We also analyze the feasibility of automating parts of evaluation with a strong LLM (GPT-4). Our framework supports a thorough assessment of LLMs as they are integrated into real-world applications. We have made publicly available the task tree, TencentLLMEval dataset, and evaluation methodology which have been demonstrated as effective in assessing the performance of Tencent Hunyuan LLMs. By doing so, we aim to facilitate the benchmarking of advances in the development of safe and human-aligned LLMs.

Large Language Model (LLM) training often optimizes for preference alignment, rewarding outputs that are perceived as helpful and interaction-friendly. However, this preference-oriented objective can be exploited: manipulative prompts can steer responses toward user-appeasing agreement and away from truth-oriented correction. In this work, we investigate whether aligned models are vulnerable to Preference-Undermining Attacks (PUA), a class of manipulative prompting strategies designed to exploit the model's desire to please user preferences at the expense of truthfulness. We propose a diagnostic methodology that provides a finer-grained and more directive analysis than aggregate benchmark scores, using a factorial evaluation framework to decompose prompt-induced shifts into interpretable effects of system objectives (truth- vs. preference-oriented) and PUA-style dialogue factors (directive control, personal derogation, conditional approval, reality denial) within a controlled 2 times 2^4 design. Surprisingly, more advanced models are sometimes more susceptible to manipulative prompts. Beyond the dominant reality-denial factor, we observe model-specific sign reversals and interactions with PUA-style factors, suggesting tailored defenses rather than uniform robustness. These findings offer a novel, reproducible factorial evaluation methodology that provides finer-grained diagnostics for post-training processes like RLHF, enabling better trade-offs in the product iteration of LLMs by offering a more nuanced understanding of preference alignment risks and the impact of manipulative prompts.

The development of audio foundation models has accelerated rapidly since the emergence of GPT-4o. However, the lack of comprehensive evaluation has become a critical bottleneck for further progress in the field, particularly in audio generation. Current audio evaluation faces three major challenges: (1) audio evaluation lacks a unified framework, with datasets and code scattered across various sources, hindering fair and efficient cross-model comparison;(2) audio codecs, as a key component of audio foundation models, lack a widely accepted and holistic evaluation methodology; (3) existing speech benchmarks are heavily reliant on English, making it challenging to objectively assess models' performance on Chinese. To address the first issue, we introduce UltraEval-Audio, a unified evaluation framework for audio foundation models, specifically designed for both audio understanding and generation tasks. UltraEval-Audio features a modular architecture, supporting 10 languages and 14 core task categories, while seamlessly integrating 24 mainstream models and 36 authoritative benchmarks. To enhance research efficiency, the framework provides a one-command evaluation feature, accompanied by real-time public leaderboards. For the second challenge, UltraEval-Audio adopts a novel comprehensive evaluation scheme for audio codecs, evaluating performance across three key dimensions: semantic accuracy, timbre fidelity, and acoustic quality. To address the third issue, we propose two new Chinese benchmarks, SpeechCMMLU and SpeechHSK, designed to assess Chinese knowledge proficiency and language fluency. We wish that UltraEval-Audio will provide both academia and industry with a transparent, efficient, and fair platform for comparison of audio models. Our code, benchmarks, and leaderboards are available at https://github.com/OpenBMB/UltraEval-Audio.

Agent benchmarks are growing fast, but no single benchmark touches more than four or five of the dimensions that deployment exposes. This paper aggregates the largest coordinated deep-dive of one MCP-based industrial-agent benchmark to date: fourteen parallel implementation studies covering new asset classes (including a multi-modal visual extension), alternative orchestrations, retrieval strategies, reasoning modes, infrastructure optimizations, and evaluation-methodology probes. Consolidating those studies with seven prior agent benchmarks, we argue that aggregate-score leaderboards systematically underspecify deployed-agent evaluation. Rankings derived from aggregate scores do not transfer to out-of-distribution settings; recent public-to-hidden competition retrospectives provide direct empirical evidence of this rank instability. We propose ranking configurations by predictive validity, the correlation between in-sample and out-of-sample rank, rather than in-sample mean, and report a twelve-tier measurement apparatus that exposes the deployment-relevant dimensions HELM and its agent-era successors collapse. The position is operationalized through three falsifiable out-of-distribution criteria with explicit thresholds; existing evidence partly supports it but is too thin to confirm. We close with a pre-registered pilot design and a field-level vision for what the next generation of agentic benchmarks should report.

Reliably extracting tables from PDFs is essential for large-scale scientific data mining and knowledge base construction, yet existing evaluation approaches rely on rule-based metrics that fail to capture semantic equivalence of table content. We present a benchmarking framework based on synthetically generated PDFs with precise LaTeX ground truth, using tables sourced from arXiv to ensure realistic complexity and diversity. As our central methodological contribution, we apply LLM-as-a-judge for semantic table evaluation, integrated into a matching pipeline that accommodates inconsistencies in parser outputs. Through a human validation study comprising over 1,500 quality judgments on extracted table pairs, we show that LLM-based evaluation achieves substantially higher correlation with human judgment (Pearson r=0.93) compared to Tree Edit Distance-based Similarity (TEDS, r=0.68) and Grid Table Similarity (GriTS, r=0.70). Evaluating 21 contemporary PDF parsers across 100 synthetic documents containing 451 tables reveals significant performance disparities. Our results offer practical guidance for selecting parsers for tabular data extraction and establish a reproducible, scalable evaluation methodology for this critical task. Code and data: https://github.com/phorn1/pdf-parse-bench Metric study and human evaluation: https://github.com/phorn1/table-metric-study

We present a scalable pipeline for automatically generating high-quality training data for web agents. In particular, a major challenge in identifying high-quality training instances is trajectory evaluation - quantifying how much progress was made towards task completion. We introduce a novel constraint-based evaluation framework that provides fine-grained assessment of progress towards task completion. This enables us to leverage partially successful trajectories, which significantly expands the amount of usable training data. We evaluate our method on a new benchmark we propose called BookingArena, which consists of complex booking tasks across 20 popular websites, and demonstrate that our distilled student model outperforms open-source approaches and matches or exceeds commercial systems, while being a significantly smaller model. Our work addresses the challenge of efficiently creating diverse, realistic web interaction datasets and provides a systematic evaluation methodology for complex structured web tasks.

Purpose: With advancements in Large Language Models (LLMs) for healthcare, the need arises for competitive open-source models to protect the public interest. This work contributes to the field of open medical LLMs by optimizing key stages of data preprocessing and training, while showing how to improve model safety (through DPO) and efficacy (through RAG). The evaluation methodology used, which includes four different types of tests, defines a new standard for the field. The resultant models, shown to be competitive with the best private alternatives, are released with a permisive license. Methods: Building on top of strong base models like Llama 3.1 and Qwen 2.5, Aloe Beta uses a custom dataset to enhance public data with synthetic Chain of Thought examples. The models undergo alignment with Direct Preference Optimization, emphasizing ethical and policy-aligned performance in the presence of jailbreaking attacks. Evaluation includes close-ended, open-ended, safety and human assessments, to maximize the reliability of results. Results: Recommendations are made across the entire pipeline, backed by the solid performance of the Aloe Family. These models deliver competitive performance across healthcare benchmarks and medical fields, and are often preferred by healthcare professionals. On bias and toxicity, the Aloe Beta models significantly improve safety, showing resilience to unseen jailbreaking attacks. For a responsible release, a detailed risk assessment specific to healthcare is attached to the Aloe Family models. Conclusion: The Aloe Beta models, and the recipe that leads to them, are a significant contribution to the open-source medical LLM field, offering top-of-the-line performance while maintaining high ethical requirements. This work sets a new standard for developing and reporting aligned LLMs in healthcare.

Multi-agent systems represent a significant advancement in artificial intelligence, enabling complex problem-solving through coordinated specialized agents. However, these systems face fundamental challenges in context management, coordination efficiency, and scalable operation. This paper introduces a comprehensive framework for advancing multi-agent systems through Model Context Protocol (MCP), addressing these challenges through standardized context sharing and coordination mechanisms. We extend previous work on AI agent architectures by developing a unified theoretical foundation, advanced context management techniques, and scalable coordination patterns. Through detailed implementation case studies across enterprise knowledge management, collaborative research, and distributed problem-solving domains, we demonstrate significant performance improvements compared to traditional approaches. Our evaluation methodology provides a systematic assessment framework with benchmark tasks and datasets specifically designed for multi-agent systems. We identify current limitations, emerging research opportunities, and potential transformative applications across industries. This work contributes to the evolution of more capable, collaborative, and context-aware artificial intelligence systems that can effectively address complex real-world challenges.

With an evergrowing number of LLMs reporting superlative performance for English, their ability to perform equitably for different dialects of English (i.e., dialect robustness) needs to be ascertained. Specifically, we use English language (US English or Indian English) conversations between humans who play the word-guessing game of `taboo'. We formulate two evaluative tasks: target word prediction (TWP) (i.e.predict the masked target word in a conversation) and target word selection (TWS) (i.e., select the most likely masked target word in a conversation, from among a set of candidate words). Extending MD3, an existing dialectic dataset of taboo-playing conversations, we introduce M-MD3, a target-word-masked version of MD3 with the USEng and IndEng subsets. We add two subsets: AITrans (where dialectic information is removed from IndEng) and AIGen (where LLMs are prompted to generate conversations). Our evaluation uses pre-trained and fine-tuned versions of two closed-source (GPT-4/3.5) and two open-source LLMs (Mistral and Gemma). LLMs perform significantly better for US English than Indian English for both TWP and TWS, for all settings. While GPT-based models perform the best, the comparatively smaller models work more equitably for short conversations (<8 turns). Our results on AIGen and AITrans (the best and worst-performing subset) respectively show that LLMs may learn a dialect of their own based on the composition of the training data, and that dialect robustness is indeed a challenging task. Our evaluation methodology exhibits a novel way to examine attributes of language models using pre-existing dialogue datasets.

Gender-bias stereotypes have recently raised significant ethical concerns in natural language processing. However, progress in detection and evaluation of gender bias in natural language understanding through inference is limited and requires further investigation. In this work, we propose an evaluation methodology to measure these biases by constructing a challenge task that involves pairing gender-neutral premises against a gender-specific hypothesis. We use our challenge task to investigate state-of-the-art NLI models on the presence of gender stereotypes using occupations. Our findings suggest that three models (BERT, RoBERTa, BART) trained on MNLI and SNLI datasets are significantly prone to gender-induced prediction errors. We also find that debiasing techniques such as augmenting the training dataset to ensure a gender-balanced dataset can help reduce such bias in certain cases.

Unsupervised speech representation learning has shown remarkable success at finding representations that correlate with phonetic structures and improve downstream speech recognition performance. However, most research has been focused on evaluating the representations in terms of their ability to improve the performance of speech recognition systems on read English (e.g. Wall Street Journal and LibriSpeech). This evaluation methodology overlooks two important desiderata that speech representations should have: robustness to domain shifts and transferability to other languages. In this paper we learn representations from up to 8000 hours of diverse and noisy speech data and evaluate the representations by looking at their robustness to domain shifts and their ability to improve recognition performance in many languages. We find that our representations confer significant robustness advantages to the resulting recognition systems: we see significant improvements in out-of-domain transfer relative to baseline feature sets and the features likewise provide improvements in 25 phonetically diverse languages including tonal languages and low-resource languages.

Advancements in LLMs have enhanced task automation in software engineering; however, current evaluations primarily focus on natural language tasks, overlooking code quality. Most benchmarks prioritize high-level reasoning over executable code and real-world performance, leaving gaps in understanding true capabilities and risks associated with these models in production. To address this issue, we propose MERA Code, a new addition to the MERA benchmark family, specifically focused on evaluating code for the latest code generation LLMs in Russian. This benchmark includes 11 evaluation tasks that span 8 programming languages. Our proposed evaluation methodology features a taxonomy that outlines the practical coding skills necessary for models to complete these tasks. The benchmark comprises an open-source codebase for users to conduct MERA assessments, a scoring system compatible with various programming environments, and a platform featuring a leaderboard and submission system. We evaluate open LLMs and frontier API models, analyzing their limitations in terms of practical coding tasks in non-English languages. We are publicly releasing MERA to guide future research, anticipate groundbreaking features in model development, and standardize evaluation procedures.

We introduce Legal RAG Bench, a benchmark and evaluation methodology for assessing the end-to-end performance of legal RAG systems. As a benchmark, Legal RAG Bench consists of 4,876 passages from the Victorian Criminal Charge Book alongside 100 complex, hand-crafted questions demanding expert knowledge of criminal law and procedure. Both long-form answers and supporting passages are provided. As an evaluation methodology, Legal RAG Bench leverages a full factorial design and novel hierarchical error decomposition framework, enabling apples-to-apples comparisons of the contributions of retrieval and reasoning models in RAG. We evaluate three state-of-the-art embedding models (Isaacus' Kanon 2 Embedder, Google's Gemini Embedding 001, and OpenAI's Text Embedding 3 Large) and two frontier LLMs (Gemini 3.1 Pro and GPT-5.2), finding that information retrieval is the primary driver of legal RAG performance, with LLMs exerting a more moderate effect on correctness and groundedness. Kanon 2 Embedder, in particular, had the largest positive impact on performance, improving average correctness by 17.5 points, groundedness by 4.5 points, and retrieval accuracy by 34 points. We observe that many errors attributed to hallucinations in legal RAG systems are in fact triggered by retrieval failures, concluding that retrieval sets the ceiling for the performance of many modern legal RAG systems. We document why and how we built Legal RAG Bench alongside the results of our evaluations. We also openly release our code and data to assist with reproduction of our findings.

We present the evaluation methodology, datasets and results of the BOP Challenge 2024, the sixth in a series of public competitions organized to capture the state of the art in 6D object pose estimation and related tasks. In 2024, our goal was to transition BOP from lab-like setups to real-world scenarios. First, we introduced new model-free tasks, where no 3D object models are available and methods need to onboard objects just from provided reference videos. Second, we defined a new, more practical 6D object detection task where identities of objects visible in a test image are not provided as input. Third, we introduced new BOP-H3 datasets recorded with high-resolution sensors and AR/VR headsets, closely resembling real-world scenarios. BOP-H3 include 3D models and onboarding videos to support both model-based and model-free tasks. Participants competed on seven challenge tracks, each defined by a task, object onboarding setup, and dataset group. Notably, the best 2024 method for model-based 6D localization of unseen objects (FreeZeV2.1) achieves 22% higher accuracy on BOP-Classic-Core than the best 2023 method (GenFlow), and is only 4% behind the best 2023 method for seen objects (GPose2023) although being significantly slower (24.9 vs 2.7s per image). A more practical 2024 method for this task is Co-op which takes only 0.8s per image and is 25X faster and 13% more accurate than GenFlow. Methods have a similar ranking on 6D detection as on 6D localization but higher run time. On model-based 2D detection of unseen objects, the best 2024 method (MUSE) achieves 21% relative improvement compared to the best 2023 method (CNOS). However, the 2D detection accuracy for unseen objects is still noticealy (-53%) behind the accuracy for seen objects (GDet2023). The online evaluation system stays open and is available at http://bop.felk.cvut.cz/

As large language models (LLMs) evolve into sophisticated autonomous agents capable of complex software development tasks, evaluating their real-world capabilities becomes critical. While existing benchmarks like LoCoBench~qiu2025locobench assess long-context code understanding, they focus on single-turn evaluation and cannot capture the multi-turn interactive nature, tool usage patterns, and adaptive reasoning required by real-world coding agents. We introduce LoCoBench-Agent, a comprehensive evaluation framework specifically designed to assess LLM agents in realistic, long-context software engineering workflows. Our framework extends LoCoBench's 8,000 scenarios into interactive agent environments, enabling systematic evaluation of multi-turn conversations, tool usage efficiency, error recovery, and architectural consistency across extended development sessions. We also introduce an evaluation methodology with 9 metrics across comprehension and efficiency dimensions. Our framework provides agents with 8 specialized tools (file operations, search, code analysis) and evaluates them across context lengths ranging from 10K to 1M tokens, enabling precise assessment of long-context performance. Through systematic evaluation of state-of-the-art models, we reveal several key findings: (1) agents exhibit remarkable long-context robustness; (2) comprehension-efficiency trade-off exists with negative correlation, where thorough exploration increases comprehension but reduces efficiency; and (3) conversation efficiency varies dramatically across models, with strategic tool usage patterns differentiating high-performing agents. As the first long-context LLM agent benchmark for software engineering, LoCoBench-Agent establishes a rigorous foundation for measuring agent capabilities, identifying performance gaps, and advancing autonomous software development at scale.

Tool-augmented LLM agents introduce security risks that extend beyond user-input filtering, including indirect prompt injection through fetched content, unsafe tool execution, credential leakage, and tampering with local control files. We present OpenClaw PRISM, a zero-fork runtime security layer for OpenClaw-based agent gateways. PRISM combines an in-process plugin with optional sidecar services and distributes enforcement across ten lifecycle hooks spanning message ingress, prompt construction, tool execution, tool-result persistence, outbound messaging, sub-agent spawning, and gateway startup. Rather than introducing a novel detection model, PRISM integrates a hybrid heuristic-plus-LLM scanning pipeline, conversation- and session-scoped risk accumulation with TTL-based decay, policy-enforced controls over tools, paths, private networks, domain tiers, and outbound secret patterns, and a tamper-evident audit and operations plane with integrity verification and hot-reloadable policy management. We outline an evaluation methodology and benchmark pipeline for measuring security effectiveness, false positives, layer contribution, runtime overhead, and operational recoverability in an agent-runtime setting, and we report current preliminary benchmark results on curated same-slice experiments and operational microbenchmarks. The system targets deployable runtime defense for real agent gateways rather than benchmark-only detection.

The prosody of a spoken utterance, including features like stress, intonation and rhythm, can significantly affect the underlying semantics, and as a consequence can also affect its textual translation. Nevertheless, prosody is rarely studied within the context of speech-to-text translation (S2TT) systems. In particular, end-to-end (E2E) systems have been proposed as well-suited for prosody-aware translation because they have direct access to the speech signal when making translation decisions, but the understanding of whether this is successful in practice is still limited. A main challenge is the difficulty of evaluating prosody awareness in translation. To address this challenge, we introduce an evaluation methodology and a focused benchmark (named ContraProST) aimed at capturing a wide range of prosodic phenomena. Our methodology uses large language models and controllable text-to-speech (TTS) to generate contrastive examples. Through experiments in translating English speech into German, Spanish, and Japanese, we find that (a) S2TT models possess some internal representation of prosody, but the prosody signal is often not strong enough to affect the translations, (b) E2E systems outperform cascades of speech recognition and text translation systems, confirming their theoretical advantage in this regard, and (c) certain cascaded systems also capture prosodic information in the translation, but only to a lesser extent that depends on the particulars of the transcript's surface form.

Efficiently tackling combinatorial reasoning problems, particularly the notorious NP-hard tasks, remains a significant challenge for AI research. Recent efforts have sought to enhance planning by incorporating hierarchical high-level search strategies, known as subgoal methods. While promising, their performance against traditional low-level planners is inconsistent, raising questions about their application contexts. In this study, we conduct an in-depth exploration of subgoal-planning methods for combinatorial reasoning. We identify the attributes pivotal for leveraging the advantages of high-level search: hard-to-learn value functions, complex action spaces, presence of dead ends in the environment, or using data collected from diverse experts. We propose a consistent evaluation methodology to achieve meaningful comparisons between methods and reevaluate the state-of-the-art algorithms.

The extraction of a small number of relevant insights from vast amounts of data is a crucial component of data-driven decision-making. However, accomplishing this task requires considerable technical skills, domain expertise, and human labor. This study explores the potential of using Large Language Models (LLMs) to automate the discovery of insights in data, leveraging recent advances in reasoning and code generation techniques. We propose a new evaluation methodology based on a "capture the flag" principle, measuring the ability of such models to recognize meaningful and pertinent information (flags) in a dataset. We further propose two proof-of-concept agents, with different inner workings, and compare their ability to capture such flags in a real-world sales dataset. While the work reported here is preliminary, our results are sufficiently interesting to mandate future exploration by the community.

In this paper, we introduce a systematic framework beyond conventional method to assess LLMs' mathematical-reasoning robustness by stress-testing them on advanced math problems that are mathematically equivalent but with linguistic and parametric variation. These transformations allow us to measure the sensitivity of LLMs to non-mathematical perturbations, thereby enabling a more accurate evaluation of their mathematical reasoning capabilities. Using this new evaluation methodology, we created PutnamGAP, a new benchmark dataset with multiple mathematically-equivalent variations of competition-level math problems. With the new dataset, we evaluate multiple families of representative LLMs and examine their robustness. Across 18 commercial and open-source models we observe sharp performance degradation on the variants. OpenAI's flagship reasoning model, O3, scores 51.5% on the originals but drops by 4.7 percentage points on surface-renaming variants, and by 12.9 percentage points on parametric variants, while smaller models fare far worse. Overall, the results show that the proposed new evaluation methodology is effective for deepening our understanding of the robustness of LLMs and generating new insights for further improving their mathematical reasoning capabilities.

Retrieval Augmented Generation (RAG) enhances the abilities of Large Language Models (LLMs) by enabling the retrieval of documents into the LLM context to provide more accurate and relevant responses. Existing RAG solutions do not focus on queries that may require fetching multiple documents with substantially different contents. Such queries occur frequently, but are challenging because the embeddings of these documents may be distant in the embedding space, making it hard to retrieve them all. This paper introduces Multi-Head RAG (MRAG), a novel scheme designed to address this gap with a simple yet powerful idea: leveraging activations of Transformer's multi-head attention layer, instead of the decoder layer, as keys for fetching multi-aspect documents. The driving motivation is that different attention heads can learn to capture different data aspects. Harnessing the corresponding activations results in embeddings that represent various facets of data items and queries, improving the retrieval accuracy for complex queries. We provide an evaluation methodology and metrics, synthetic datasets, and real-world use cases to demonstrate MRAG's effectiveness, showing improvements of up to 20% in relevance over standard RAG baselines. MRAG can be seamlessly integrated with existing RAG frameworks and benchmarking tools like RAGAS as well as different classes of data stores.

We present the evaluation methodology, datasets and results of the BOP Challenge 2023, the fifth in a series of public competitions organized to capture the state of the art in model-based 6D object pose estimation from an RGB/RGB-D image and related tasks. Besides the three tasks from 2022 (model-based 2D detection, 2D segmentation, and 6D localization of objects seen during training), the 2023 challenge introduced new variants of these tasks focused on objects unseen during training. In the new tasks, methods were required to learn new objects during a short onboarding stage (max 5 minutes, 1 GPU) from provided 3D object models. The best 2023 method for 6D localization of unseen objects (GenFlow) notably reached the accuracy of the best 2020 method for seen objects (CosyPose), although being noticeably slower. The best 2023 method for seen objects (GPose) achieved a moderate accuracy improvement but a significant 43% run-time improvement compared to the best 2022 counterpart (GDRNPP). Since 2017, the accuracy of 6D localization of seen objects has improved by more than 50% (from 56.9 to 85.6 AR_C). The online evaluation system stays open and is available at: http://bop.felk.cvut.cz/.

Event cameras are a new type of vision sensor that incorporates asynchronous and independent pixels, offering advantages over traditional frame-based cameras such as high dynamic range and minimal motion blur. However, their output is not easily understandable by humans, making the reconstruction of intensity images from event streams a fundamental task in event-based vision. While recent deep learning-based methods have shown promise in video reconstruction from events, this problem is not completely solved yet. To facilitate comparison between different approaches, standardized evaluation protocols and diverse test datasets are essential. This paper proposes a unified evaluation methodology and introduces an open-source framework called EVREAL to comprehensively benchmark and analyze various event-based video reconstruction methods from the literature. Using EVREAL, we give a detailed analysis of the state-of-the-art methods for event-based video reconstruction, and provide valuable insights into the performance of these methods under varying settings, challenging scenarios, and downstream tasks.

We present the evaluation methodology, datasets and results of the BOP Challenge 2022, the fourth in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB/RGB-D image. In 2022, we witnessed another significant improvement in the pose estimation accuracy -- the state of the art, which was 56.9 AR_C in 2019 (Vidal et al.) and 69.8 AR_C in 2020 (CosyPose), moved to new heights of 83.7 AR_C (GDRNPP). Out of 49 pose estimation methods evaluated since 2019, the top 18 are from 2022. Methods based on point pair features, which were introduced in 2010 and achieved competitive results even in 2020, are now clearly outperformed by deep learning methods. The synthetic-to-real domain gap was again significantly reduced, with 82.7 AR_C achieved by GDRNPP trained only on synthetic images from BlenderProc. The fastest variant of GDRNPP reached 80.5 AR_C with an average time per image of 0.23s. Since most of the recent methods for 6D object pose estimation begin by detecting/segmenting objects, we also started evaluating 2D object detection and segmentation performance based on the COCO metrics. Compared to the Mask R-CNN results from CosyPose in 2020, detection improved from 60.3 to 77.3 AP_C and segmentation from 40.5 to 58.7 AP_C. The online evaluation system stays open and is available at: http://bop.felk.cvut.cz/{bop.felk.cvut.cz}.

Opinion and sentiment analysis is a vital task to characterize subjective information in social media posts. In this paper, we present a comprehensive experimental evaluation and comparison with six state-of-the-art methods, from which we have re-implemented one of them. In addition, we investigate different textual and visual feature embeddings that cover different aspects of the content, as well as the recently introduced multimodal CLIP embeddings. Experimental results are presented for two different publicly available benchmark datasets of tweets and corresponding images. In contrast to the evaluation methodology of previous work, we introduce a reproducible and fair evaluation scheme to make results comparable. Finally, we conduct an error analysis to outline the limitations of the methods and possibilities for the future work.

Supervised machine learning provides the learner with a set of input-output examples of the target task. Humans, however, can also learn to perform new tasks from instructions in natural language. Can machines learn to understand instructions as well? We present the Turking Test, which examines a model's ability to follow natural language instructions of varying complexity. These range from simple tasks, like retrieving the nth word of a sentence, to ones that require creativity, such as generating examples for SNLI and SQuAD in place of human intelligence workers ("turkers"). Despite our lenient evaluation methodology, we observe that a large pretrained language model performs poorly across all tasks. Analyzing the model's error patterns reveals that the model tends to ignore explicit instructions and often generates outputs that cannot be construed as an attempt to solve the task. While it is not yet clear whether instruction understanding can be captured by traditional language models, the sheer expressivity of instruction understanding makes it an appealing alternative to the rising few-shot inference paradigm.

This paper presents the evaluation methodology, datasets, and results of the BOP Challenge 2020, the third in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB-D image. In 2020, to reduce the domain gap between synthetic training and real test RGB images, the participants were provided 350K photorealistic training images generated by BlenderProc4BOP, a new open-source and light-weight physically-based renderer (PBR) and procedural data generator. Methods based on deep neural networks have finally caught up with methods based on point pair features, which were dominating previous editions of the challenge. Although the top-performing methods rely on RGB-D image channels, strong results were achieved when only RGB channels were used at both training and test time - out of the 26 evaluated methods, the third method was trained on RGB channels of PBR and real images, while the fifth on RGB channels of PBR images only. Strong data augmentation was identified as a key component of the top-performing CosyPose method, and the photorealism of PBR images was demonstrated effective despite the augmentation. The online evaluation system stays open and is available on the project website: bop.felk.cvut.cz.

Recent large language models (LLMs) have demonstrated great potential toward intelligent agents and next-gen automation, but there currently lacks a systematic benchmark for evaluating LLMs' abilities as agents. We introduce SmartPlay: both a challenging benchmark and a methodology for evaluating LLMs as agents. SmartPlay consists of 6 different games, including Rock-Paper-Scissors, Tower of Hanoi, Minecraft. Each game features a unique setting, providing up to 20 evaluation settings and infinite environment variations. Each game in SmartPlay uniquely challenges a subset of 9 important capabilities of an intelligent LLM agent, including reasoning with object dependencies, planning ahead, spatial reasoning, learning from history, and understanding randomness. The distinction between the set of capabilities each game test allows us to analyze each capability separately. SmartPlay serves not only as a rigorous testing ground for evaluating the overall performance of LLM agents but also as a road-map for identifying gaps in current methodologies. We release our benchmark at github.com/LLMsmartplay/SmartPlay

Taiwanese Hokkien (Taigi) presents unique opportunities for advancing speech technology methodologies that can generalize to diverse linguistic contexts. We introduce Breeze Taigi, a comprehensive framework centered on standardized benchmarks for evaluating Taigi speech recognition and synthesis systems. Our primary contribution is a reproducible evaluation methodology that leverages parallel Taiwanese Mandarin resources. We provide 30 carefully curated Mandarin-Taigi audio pairs from Taiwan's Executive Yuan public service announcements with normalized ground truth transcriptions. We establish Character Error Rate (CER) as the standard metric and implement normalization procedures to enable fair cross-system comparisons. To demonstrate the benchmark's utility and provide reference implementations, we develop speech recognition and synthesis models through a methodology that leverages existing Taiwanese Mandarin resources and large-scale synthetic data generation. In particular, we fine-tune a Whisper model on approximately 10,000 hours of Taigi synthetic speech data. Our ASR model achieves 30.13% average CER on the benchmark, outperforming existing commercial and research systems. By providing standardized evaluation protocols, diverse training datasets, and open baseline models, we offer a replicable framework with methodologies applicable to various linguistic contexts.

Using offline datasets to evaluate conversational agents often fails to cover rare scenarios or to support testing new policies. This has motivated the use of controllable user simulators for targeted, counterfactual evaluation, typically implemented by prompting or fine-tuning large language models. In this work, we formalize controllable simulation as a causal inference problem. By bridging natural language evaluation with off-policy evaluation methodology, we show that the standard practice of training simulators via supervised fine-tuning on post-hoc trajectory labels yields a structurally biased model. Specifically, these labels are inextricably coupled to the data-generating behavior policy, injecting a look-ahead bias that breaks causal consistency. Furthermore, we prove that under policy shift this failure causes the variance of evaluation metrics to explode geometrically, a phenomenon we term controllability collapse. To restore causal consistency, we establish theoretical conditions for accurate simulation and propose practical training mitigations: a priori controls, step-wise dynamic controls, and direct policy-conditioned learning. Empirical evaluation confirms that while standard global controls distort conversational distributions and collapse behavioral diversity, our causally grounded simulators eliminate look-ahead bias, preserve natural variance, and exhibit robust zero-shot generalization to unseen agent behaviors.

Large Language Models (LLMs) have achieved remarkable progress on advanced reasoning tasks such as mathematics and coding competitions. Meanwhile, physics, despite being both reasoning-intensive and essential to real-world understanding, received limited academic and industrial attention. This paper introduces PHYSICS, a dataset containing 16,568 high-quality physics problems spanning subjects and difficulty levels, to facilitate this issue. Specifically, PHYSICS is curated with exercises from over 100 textbooks through a carefully designed pipeline for quality control. It covers five major physics domains: Mechanics, Electromagnetism, Thermodynamics, Optics, and Modern Physics. It also spans a wide range of difficulty levels, from high school to graduate-level physics courses. To utilize the data for improving and evaluating the model's physical reasoning capabilities, we split the dataset into training and test sets, and provide reasoning paths generated by powerful reasoning models for the training data to facilitate model training. In addition, for the evaluation part, we find that existing evaluation frameworks exhibit biases in aspects such as units, simplification, and precision in physics domain. To balance efficiency and accuracy, we introduce a Rule+Model evaluation framework tailored to physics problems. Our evaluations on current state-of-the-art open-source and proprietary models highlight the limitations of current models in handling physics-related tasks. We hope that our dataset and evaluation methodology will jointly advance the development of LLMs in the field of physics.

Many companies use large language models (LLMs) offered as a service, like OpenAI's GPT-4, to create AI-enabled product experiences. Along with the benefits of ease-of-use and shortened time-to-solution, this reliance on proprietary services has downsides in model control, performance reliability, uptime predictability, and cost. At the same time, a flurry of open-source small language models (SLMs) has been made available for commercial use. However, their readiness to replace existing capabilities remains unclear, and a systematic approach to holistically evaluate these SLMs is not readily available. This paper presents a systematic evaluation methodology and a characterization of modern open-source SLMs and their trade-offs when replacing proprietary LLMs for a real-world product feature. We have designed SLaM, an open-source automated analysis tool that enables the quantitative and qualitative testing of product features utilizing arbitrary SLMs. Using SLaM, we examine the quality and performance characteristics of modern SLMs relative to an existing customer-facing implementation using the OpenAI GPT-4 API. Across 9 SLMs and their 29 variants, we observe that SLMs provide competitive results, significant performance consistency improvements, and a cost reduction of 5x~29x when compared to GPT-4.

Neural information retrieval (IR) systems have progressed rapidly in recent years, in large part due to the release of publicly available benchmarking tasks. Unfortunately, some dimensions of this progress are illusory: the majority of the popular IR benchmarks today focus exclusively on downstream task accuracy and thus conceal the costs incurred by systems that trade away efficiency for quality. Latency, hardware cost, and other efficiency considerations are paramount to the deployment of IR systems in user-facing settings. We propose that IR benchmarks structure their evaluation methodology to include not only metrics of accuracy, but also efficiency considerations such as a query latency and the corresponding cost budget for a reproducible hardware setting. For the popular IR benchmarks MS MARCO and XOR-TyDi, we show how the best choice of IR system varies according to how these efficiency considerations are chosen and weighed. We hope that future benchmarks will adopt these guidelines toward more holistic IR evaluation.

We introduce GODEL (Grounded Open Dialogue Language Model), a large pre-trained language model for dialog. In contrast with earlier models such as DialoGPT, GODEL leverages a new phase of grounded pre-training designed to better support adapting GODEL to a wide range of downstream dialog tasks that require information external to the current conversation (e.g., a database or document) to produce good responses. Experiments against an array of benchmarks that encompass task-oriented dialog, conversational QA, and grounded open-domain dialog show that GODEL outperforms state-of-the-art pre-trained dialog models in few-shot fine-tuning setups, in terms of both human and automatic evaluation. A novel feature of our evaluation methodology is the introduction of a notion of utility that assesses the usefulness of responses (extrinsic evaluation) in addition to their communicative features (intrinsic evaluation). We show that extrinsic evaluation offers improved inter-annotator agreement and correlation with automated metrics. Code and data processing scripts are publicly available.

Single-prompt accuracy is the dominant way to benchmark language models, but it can miss reliability failures that matter. We evaluate a 15-model open-weight corpus, with the main reliability analyses focused on 10 instruct models across five classification and reasoning benchmarks under five prompt variants each, measuring accuracy, token-probability calibration, verbal-confidence calibration, verbal parse rate, and prompt-perturbation spread for every (model x dataset x variant) cell. We find three broad results. First, evaluation design can materially change the conclusion. Switching Expected Calibration Error (ECE) token from a raw to a label-set-normalised definition changes per-cell calibration by a mean absolute 0.149. More strikingly, pairing a chain-of-thought prompt with a first-character evaluator on ARC-Challenge reduces apparent accuracy by 72-88% across all five primary models; two independent repair procedures recover 93.8% and 102.7% of the lost performance, indicating an evaluator-side rather than model-side failure. Second, confidence signals are fragile. On MMLU-Pro, every primary model verbally reports confidence substantially above both its accuracy and its token-probability confidence on the same rows, and verbal parse rate can collapse for a single model on a single prompt variant. Third, prompt robustness does not track parameter count reliably. Across 10 instruct models, the correlation between model size and prompt-perturbation spread ranges from -0.244 to 0.474 across benchmarks. Taken together, these results show that reliability conclusions for small language models depend not only on the model being evaluated, but also on the evaluation pipeline used to measure it. We argue that calibration definitions, evaluator logic, verbal parseability, and prompt robustness should be reported explicitly when making reliability claims.

Automated neural network architecture design remains a significant challenge in computer vision. Task diversity and computational constraints require both effective architectures and efficient search methods. Large Language Models (LLMs) present a promising alternative to computationally intensive Neural Architecture Search (NAS), but their application to architecture generation in computer vision has not been systematically studied, particularly regarding prompt engineering and validation strategies. Building on the task-agnostic NNGPT/LEMUR framework, this work introduces and validates two key contributions for computer vision. First, we present Few-Shot Architecture Prompting (FSAP), the first systematic study of the number of supporting examples (n = 1, 2, 3, 4, 5, 6) for LLM-based architecture generation. We find that using n = 3 examples best balances architectural diversity and context focus for vision tasks. Second, we introduce Whitespace-Normalized Hash Validation, a lightweight deduplication method (less than 1 ms) that provides a 100x speedup over AST parsing and prevents redundant training of duplicate computer vision architectures. In large-scale experiments across seven computer vision benchmarks (MNIST, CIFAR-10, CIFAR-100, CelebA, ImageNette, SVHN, Places365), we generated 1,900 unique architectures. We also introduce a dataset-balanced evaluation methodology to address the challenge of comparing architectures across heterogeneous vision tasks. These contributions provide actionable guidelines for LLM-based architecture search in computer vision and establish rigorous evaluation practices, making automated design more accessible to researchers with limited computational resources.

This paper introduces the first publicly available dataset for Automatic Essay Scoring (AES) and feedback generation in Basque, targeting the CEFR C1 proficiency level. The dataset comprises 3,200 essays from HABE, each annotated by expert evaluators with criterion specific scores covering correctness, richness, coherence, cohesion, and task alignment enriched with detailed feedback and error examples. We fine-tune open-source models, including RoBERTa-EusCrawl and Latxa 8B/70B, for both scoring and explanation generation. Our experiments show that encoder models remain highly reliable for AES, while supervised fine-tuning (SFT) of Latxa significantly enhances performance, surpassing state-of-the-art (SoTA) closed-source systems such as GPT-5 and Claude Sonnet 4.5 in scoring consistency and feedback quality. We also propose a novel evaluation methodology for assessing feedback generation, combining automatic consistency metrics with expert-based validation of extracted learner errors. Results demonstrate that the fine-tuned Latxa model produces criterion-aligned, pedagogically meaningful feedback and identifies a wider range of error types than proprietary models. This resource and benchmark establish a foundation for transparent, reproducible, and educationally grounded NLP research in low-resource languages such as Basque.

Bias in text-to-image (T2I) models can propagate unfair social representations and may be used to aggressively market ideas or push controversial agendas. Existing T2I model bias evaluation methods only focus on social biases. We look beyond that and instead propose an evaluation methodology to quantify general biases in T2I generative models, without any preconceived notions. We assess four state-of-the-art T2I models and compare their baseline bias characteristics to their respective variants (two for each), where certain biases have been intentionally induced. We propose three evaluation metrics to assess model biases including: (i) Distribution bias, (ii) Jaccard hallucination and (iii) Generative miss-rate. We conduct two evaluation studies, modelling biases under general, and task-oriented conditions, using a marketing scenario as the domain for the latter. We also quantify social biases to compare our findings to related works. Finally, our methodology is transferred to evaluate captioned-image datasets and measure their bias. Our approach is objective, domain-agnostic and consistently measures different forms of T2I model biases. We have developed a web application and practical implementation of what has been proposed in this work, which is at https://huggingface.co/spaces/JVice/try-before-you-bias. A video series with demonstrations is available at https://www.youtube.com/channel/UCk-0xyUyT0MSd_hkp4jQt1Q

Language serves as a vehicle for conveying thought, enabling communication among individuals. The ability to distinguish between diverse concepts, identify fairness and injustice, and comprehend a range of legal notions fundamentally relies on logical reasoning. Large Language Models (LLMs) attempt to emulate human language understanding and generation, but their competency in logical reasoning remains limited. This paper seeks to address the philosophical question: How can we effectively teach logical reasoning to LLMs while maintaining a deep understanding of the intricate relationship between language and logic? By focusing on bolstering LLMs' capabilities in logical reasoning, we aim to expand their applicability in law and other logic-intensive disciplines. To this end, we propose a Reinforcement Learning from Logical Feedback (RLLF) approach, which serves as a potential framework for refining LLMs' reasoning capacities. Through RLLF and a revised evaluation methodology, we explore new avenues for research in this domain and contribute to the development of LLMs capable of handling complex legal reasoning tasks while acknowledging the fundamental connection between language and logic.

This paper presents the first few-shot LLM-based chatbot that almost never hallucinates and has high conversationality and low latency. WikiChat is grounded on the English Wikipedia, the largest curated free-text corpus. WikiChat generates a response from an LLM, retains only the grounded facts, and combines them with additional information it retrieves from the corpus to form factual and engaging responses. We distill WikiChat based on GPT-4 into a 7B-parameter LLaMA model with minimal loss of quality, to significantly improve its latency, cost and privacy, and facilitate research and deployment. Using a novel hybrid human-and-LLM evaluation methodology, we show that our best system achieves 97.3% factual accuracy in simulated conversations. It significantly outperforms all retrieval-based and LLM-based baselines, and by 3.9%, 38.6% and 51.0% on head, tail and recent knowledge compared to GPT-4. Compared to previous state-of-the-art retrieval-based chatbots, WikiChat is also significantly more informative and engaging, just like an LLM. WikiChat achieves 97.9% factual accuracy in conversations with human users about recent topics, 55.0% better than GPT-4, while receiving significantly higher user ratings and more favorable comments.

Large Language Models (LLMs) have demonstrated significant potential in decision-making and reasoning, particularly when integrated with various tools to effectively solve complex problems. However, existing benchmarks for evaluating LLMs' tool usage face several limitations: (1) limited evaluation scenarios, often lacking assessments in real multi-turn dialogue contexts; (2) narrow evaluation dimensions, with insufficient detailed assessments of how LLMs use tools; and (3) reliance on LLMs or real API executions for evaluation, which introduces significant overhead. To address these challenges, we introduce ACEBench, a comprehensive benchmark for assessing tool usage in LLMs. ACEBench categorizes data into three primary types based on evaluation methodology: Normal, Special, and Agent. "Normal" evaluates tool usage in basic scenarios; "Special" evaluates tool usage in situations with ambiguous or incomplete instructions; "Agent" evaluates tool usage through multi-agent interactions to simulate real-world, multi-turn dialogues. We conducted extensive experiments using ACEBench, analyzing various LLMs in-depth and providing a more granular examination of error causes across different data types.

Current large language model (LLM) development treats task-solving and preference alignment as separate challenges, optimizing first for objective correctness, then for alignment to aggregated human preferences. This paradigm fails in human-facing applications where solving a problem correctly is insufficient if the response mismatches the user's needs. This challenge intensifies in just-in-time scenarios where no prior user interaction history exists due to cold-start conditions or privacy constraints. LLMs need to identify what they don't know about user preferences, strategically elicit preference values through questioning, then adapt their reasoning processes and responses accordingly -- a complicated chain of cognitive processes which we term personalized reasoning. We introduce PREFDISCO, an evaluation methodology that transforms static benchmarks into interactive personalization tasks using psychologically-grounded personas with sparse preferences. Our framework creates scenarios where identical questions require different reasoning chains depending on user context, as optimal explanation approaches vary by individual expertise and preferences while maintaining factual accuracy. Evaluation of 21 frontier models across 10 tasks reveals 29.0% of naive personalization attempts produce worse preference alignment than generic responses, yet generic responses also fail to serve individual user needs effectively. These findings suggest personalized reasoning requires dedicated development rather than emerging naturally. PREFDISCO establishes personalized reasoning as a measurable research frontier and reveals fundamental limitations in current LLMs' interactive capabilities, providing a foundation for developing systems that can adapt to individual users in education, healthcare, and technical domains where personalization is critical.

Tasks that require character-level reasoning, such as counting or locating characters within words, remain challenging for contemporary language models. A common conjecture is that language models' reliance on subword units, rather than characters, contributes to their struggles with character-level tasks, yet recent studies offer conflicting conclusions about the role of tokenization, leaving its impact unclear. To address this gap, we introduce CharBench, a comprehensive benchmark of character-level tasks that is two orders of magnitude larger than existing alternatives. We evaluate a diverse range of leading open-weight and proprietary models on CharBench and find that it presents a significant challenge to modern LLMs, with an average accuracy of 43.6% and 32.3% on some tasks. We present an in-depth analysis of how intrinsic properties of words and their segmentations into tokens correspond to model performance. For counting tasks, we find that tokenization properties are weakly correlated with correctness, while the length of the queried word and the actual character count play a more significant part. In contrast, for tasks requiring intra-word positional understanding, performance is negatively correlated with the length of the token containing the queried character, suggesting that longer tokens obscure character position information for LLMs. We encourage future work to build on the benchmark and evaluation methodology introduced here as tools for improving model performance on such tasks.

This paper investigates the role of dynamic external knowledge integration in improving counter-argument generation using Large Language Models (LLMs). While LLMs have shown promise in argumentative tasks, their tendency to generate lengthy, potentially unfactual responses highlights the need for more controlled and evidence-based approaches. We introduce a new manually curated dataset of argument and counter-argument pairs specifically designed to balance argumentative complexity with evaluative feasibility. We also propose a new LLM-as-a-Judge evaluation methodology that shows a stronger correlation with human judgments compared to traditional reference-based metrics. Our experimental results demonstrate that integrating dynamic external knowledge from the web significantly improves the quality of generated counter-arguments, particularly in terms of relatedness, persuasiveness, and factuality. The findings suggest that combining LLMs with real-time external knowledge retrieval offers a promising direction for developing more effective and reliable counter-argumentation systems.

Hierarchical text classification (HTC) is the task of assigning labels to a text within a structured space organized as a hierarchy. Recent works treat HTC as a conventional multilabel classification problem, therefore evaluating it as such. We instead propose to evaluate models based on specifically designed hierarchical metrics and we demonstrate the intricacy of metric choice and prediction inference method. We introduce a new challenging dataset and we evaluate fairly, recent sophisticated models, comparing them with a range of simple but strong baselines, including a new theoretically motivated loss. Finally, we show that those baselines are very often competitive with the latest models. This highlights the importance of carefully considering the evaluation methodology when proposing new methods for HTC. Code implementation and dataset are available at https://github.com/RomanPlaud/revisitingHTC.

This work introduces Salamandra, a suite of open-source decoder-only large language models available in three different sizes: 2, 7, and 40 billion parameters. The models were trained from scratch on highly multilingual data that comprises text in 35 European languages and code. Our carefully curated corpus is made exclusively from open-access data compiled from a wide variety of sources. Along with the base models, supplementary checkpoints that were fine-tuned on public-domain instruction data are also released for chat applications. Additionally, we also share our preliminary experiments on multimodality, which serve as proof-of-concept to showcase potential applications for the Salamandra family. Our extensive evaluations on multilingual benchmarks reveal that Salamandra has strong capabilities, achieving competitive performance when compared to similarly sized open-source models. We provide comprehensive evaluation results both on standard downstream tasks as well as key aspects related to bias and safety.With this technical report, we intend to promote open science by sharing all the details behind our design choices, data curation strategy and evaluation methodology. In addition to that, we deviate from the usual practice by making our training and evaluation scripts publicly accessible. We release all models under a permissive Apache 2.0 license in order to foster future research and facilitate commercial use, thereby contributing to the open-source ecosystem of large language models.

The proliferation of Large Language Models (LLMs) presents transformative potential for healthcare, yet practical deployment is hindered by the absence of frameworks that assess real-world clinical utility. Existing benchmarks test static knowledge, failing to capture the dynamic, application-oriented capabilities required in clinical practice. To bridge this gap, we introduce a Medical LLM Benchmark MLB, a comprehensive benchmark evaluating LLMs on both foundational knowledge and scenario-based reasoning. MLB is structured around five core dimensions: Medical Knowledge (MedKQA), Safety and Ethics (MedSE), Medical Record Understanding (MedRU), Smart Services (SmartServ), and Smart Healthcare (SmartCare). The benchmark integrates 22 datasets (17 newly curated) from diverse Chinese clinical sources, covering 64 clinical specialties. Its design features a rigorous curation pipeline involving 300 licensed physicians. Besides, we provide a scalable evaluation methodology, centered on a specialized judge model trained via Supervised Fine-Tuning (SFT) on expert annotations. Our comprehensive evaluation of 10 leading models reveals a critical translational gap: while the top-ranked model, Kimi-K2-Instruct (77.3% accuracy overall), excels in structured tasks like information extraction (87.8% accuracy in MedRU), performance plummets in patient-facing scenarios (61.3% in SmartServ). Moreover, the exceptional safety score (90.6% in MedSE) of the much smaller Baichuan-M2-32B highlights that targeted training is equally critical. Our specialized judge model, trained via SFT on a 19k expert-annotated medical dataset, achieves 92.1% accuracy, an F1-score of 94.37%, and a Cohen's Kappa of 81.3% for human-AI consistency, validating a reproducible and expert-aligned evaluation protocol. MLB thus provides a rigorous framework to guide the development of clinically viable LLMs.

Open-domain Knowledge Graph Completion (KGC) faces significant challenges in an ever-changing world, especially when considering the continual emergence of new entities in daily news. Existing approaches for KGC mainly rely on pretrained language models' parametric knowledge, pre-constructed queries, or single-step retrieval, typically requiring substantial supervision and training data. Even so, they often fail to capture comprehensive and up-to-date information about unpopular and/or emerging entities. To this end, we introduce Agentic Reasoning for Emerging Entities (AgREE), a novel agent-based framework that combines iterative retrieval actions and multi-step reasoning to dynamically construct rich knowledge graph triplets. Experiments show that, despite requiring zero training efforts, AgREE significantly outperforms existing methods in constructing knowledge graph triplets, especially for emerging entities that were not seen during language models' training processes, outperforming previous methods by up to 13.7%. Moreover, we propose a new evaluation methodology that addresses a fundamental weakness of existing setups and a new benchmark for KGC on emerging entities. Our work demonstrates the effectiveness of combining agent-based reasoning with strategic information retrieval for maintaining up-to-date knowledge graphs in dynamic information environments.

Structured outputs are essential for large language models (LLMs) in critical applications like agents and information extraction. Despite their capabilities, LLMs often generate outputs that deviate from predefined schemas, significantly hampering reliable application development. We present SLOT (Structured LLM Output Transformer), a model-agnostic approach that transforms unstructured LLM outputs into precise structured formats. While existing solutions predominantly rely on constrained decoding techniques or are tightly coupled with specific models, SLOT employs a fine-tuned lightweight language model as a post-processing layer, achieving flexibility across various LLMs and schema specifications. We introduce a systematic pipeline for data curation and synthesis alongside a formal evaluation methodology that quantifies both schema accuracy and content fidelity. Our results demonstrate that fine-tuned Mistral-7B model with constrained decoding achieves near perfect schema accuracy (99.5%) and content similarity (94.0%), outperforming Claude-3.5-Sonnet by substantial margins (+25 and +20 percentage points, respectively). Notably, even compact models like Llama-3.2-1B can match or exceed the structured output capabilities of much larger proprietary models when equipped with SLOT, enabling reliable structured generation in resource-constrained environments.

A critical enabler for progress in neuromorphic computing research is the ability to transparently evaluate different neuromorphic solutions on important tasks and to compare them to state-of-the-art conventional solutions. The Intel Neuromorphic Deep Noise Suppression Challenge (Intel N-DNS Challenge), inspired by the Microsoft DNS Challenge, tackles a ubiquitous and commercially relevant task: real-time audio denoising. Audio denoising is likely to reap the benefits of neuromorphic computing due to its low-bandwidth, temporal nature and its relevance for low-power devices. The Intel N-DNS Challenge consists of two tracks: a simulation-based algorithmic track to encourage algorithmic innovation, and a neuromorphic hardware (Loihi 2) track to rigorously evaluate solutions. For both tracks, we specify an evaluation methodology based on energy, latency, and resource consumption in addition to output audio quality. We make the Intel N-DNS Challenge dataset scripts and evaluation code freely accessible, encourage community participation with monetary prizes, and release a neuromorphic baseline solution which shows promising audio quality, high power efficiency, and low resource consumption when compared to Microsoft NsNet2 and a proprietary Intel denoising model used in production. We hope the Intel N-DNS Challenge will hasten innovation in neuromorphic algorithms research, especially in the area of training tools and methods for real-time signal processing. We expect the winners of the challenge will demonstrate that for problems like audio denoising, significant gains in power and resources can be realized on neuromorphic devices available today compared to conventional state-of-the-art solutions.

This paper reports on the 2018 PIRM challenge on perceptual super-resolution (SR), held in conjunction with the Perceptual Image Restoration and Manipulation (PIRM) workshop at ECCV 2018. In contrast to previous SR challenges, our evaluation methodology jointly quantifies accuracy and perceptual quality, therefore enabling perceptual-driven methods to compete alongside algorithms that target PSNR maximization. Twenty-one participating teams introduced algorithms which well-improved upon the existing state-of-the-art methods in perceptual SR, as confirmed by a human opinion study. We also analyze popular image quality measures and draw conclusions regarding which of them correlates best with human opinion scores. We conclude with an analysis of the current trends in perceptual SR, as reflected from the leading submissions.

Voice-based conversational AI systems increasingly rely on cascaded architectures that combine speech-to-text (STT), large language models (LLMs), and text-to-speech (TTS) components. We present a large-scale empirical comparison of STT x LLM x TTS stacks using data sampled from over 300,000 AI-conducted job interviews. We used an LLM-as-a-Judge automated evaluation framework to assess conversational quality, technical accuracy, and skill assessment capabilities. Our analysis of five production configurations reveals that a stack combining Google's STT, GPT-4.1, and Cartesia's TTS outperforms alternatives in both objective quality metrics and user satisfaction scores. Surprisingly, we find that objective quality metrics correlate weakly with user satisfaction scores, suggesting that user experience in voice-based AI systems depends on factors beyond technical performance. Our findings provide practical guidance for selecting components in multimodal conversations and contribute a validated evaluation methodology for human-AI interactions.

Retrieval Augmented Generation (RAG) has emerged as a standard paradigm for enhancing the factual accuracy and contextual relevance of Large Language Models (LLMs) by integrating retrieval mechanisms. However, existing evaluation frameworks fail to provide a holistic black-box approach to assessing RAG systems, especially in real-world deployment scenarios. To address this gap, we introduce SCARF (System for Comprehensive Assessment of RAG Frameworks), a modular and flexible evaluation framework designed to benchmark deployed RAG applications systematically. SCARF provides an end-to-end, black-box evaluation methodology, enabling a limited-effort comparison across diverse RAG frameworks. Our framework supports multiple deployment configurations and facilitates automated testing across vector databases and LLM serving strategies, producing a detailed performance report. Moreover, SCARF integrates practical considerations such as response coherence, providing a scalable and adaptable solution for researchers and industry professionals evaluating RAG applications. Using the REST APIs interface, we demonstrate how SCARF can be applied to real-world scenarios, showcasing its flexibility in assessing different RAG frameworks and configurations. SCARF is available at GitHub repository.

This paper gives an overview of the Schema-Guided Dialogue State Tracking task of the 8th Dialogue System Technology Challenge. The goal of this task is to develop dialogue state tracking models suitable for large-scale virtual assistants, with a focus on data-efficient joint modeling across domains and zero-shot generalization to new APIs. This task provided a new dataset consisting of over 16000 dialogues in the training set spanning 16 domains to highlight these challenges, and a baseline model capable of zero-shot generalization to new APIs. Twenty-five teams participated, developing a range of neural network models, exceeding the performance of the baseline model by a very high margin. The submissions incorporated a variety of pre-trained encoders and data augmentation techniques. This paper describes the task definition, dataset and evaluation methodology. We also summarize the approach and results of the submitted systems to highlight the overall trends in the state-of-the-art.

World models enable agents to anticipate the effects of their actions by internalizing environment dynamics. In enterprise systems, however, these dynamics are often defined by tenant-specific business logic that varies across deployments and evolves over time, making models trained on historical transitions brittle under deployment shift. We ask a question the world-models literature has not addressed: when the rules can be read at inference time, does an agent still need to learn them? We argue, and demonstrate empirically, that in settings where transition dynamics are configurable and readable, runtime discovery complements offline training by grounding predictions in the active system instance. We propose enterprise discovery agents, which recover relevant transition dynamics at runtime by reading the system's configuration rather than relying solely on internalized representations. We introduce CascadeBench, a reasoning-focused benchmark for enterprise cascade prediction that adopts the evaluation methodology of World of Workflows on diverse synthetic environments, and use it together with deployment-shift evaluation to show that offline-trained world models can perform well in-distribution but degrade as dynamics change, whereas discovery-based agents are more robust under shift by grounding their predictions in the current instance. Our findings suggest that, in configurable enterprise environments, agents should not rely solely on fixed internalized dynamics, but should incorporate mechanisms for discovering relevant transition logic at runtime.

In an era where visual content generation is increasingly driven by machine learning, the integration of human feedback into generative models presents significant opportunities for enhancing user experience and output quality. This study explores strategies for incorporating iterative human feedback into the generative process of diffusion-based text-to-image models. We propose FABRIC, a training-free approach applicable to a wide range of popular diffusion models, which exploits the self-attention layer present in the most widely used architectures to condition the diffusion process on a set of feedback images. To ensure a rigorous assessment of our approach, we introduce a comprehensive evaluation methodology, offering a robust mechanism to quantify the performance of generative visual models that integrate human feedback. We show that generation results improve over multiple rounds of iterative feedback through exhaustive analysis, implicitly optimizing arbitrary user preferences. The potential applications of these findings extend to fields such as personalized content creation and customization.

Code-switching (CSW), the alternation of languages and scripts within a single utterance, remains a fundamental challenge for multiling ual NLP, even amidst the rapid advances of large language models (LLMs). Most LLMs still struggle with mixed-language inputs, limited CSW datasets, and evaluation biases, hindering deployment in multilingual societies. This survey provides the first comprehensive analysis of CSW-aware LLM research, reviewing unique_references studies spanning five research areas, 12 NLP tasks, 30+ datasets, and 80+ languages. We classify recent advances by architecture, training strategy, and evaluation methodology, outlining how LLMs have reshaped CSW modeling and what challenges persist. The paper concludes with a roadmap emphasizing the need for inclusive datasets, fair evaluation, and linguistically grounded models to achieve truly multilingual intelligence. A curated collection of all resources is maintained at https://github.com/lingo-iitgn/awesome-code-mixing/.

The integration of Large Language Models (LLMs) with Internet-of-Things (IoT) systems faces significant challenges in hardware heterogeneity and control complexity. The Model Context Protocol (MCP) emerges as a critical enabler, providing standardized communication between LLMs and physical devices. We propose IoT-MCP, a novel framework that implements MCP through edge-deployed servers to bridge LLMs and IoT ecosystems. To support rigorous evaluation, we introduce IoT-MCP Bench, the first benchmark containing 114 Basic Tasks (e.g., ``What is the current temperature?'') and 1,140 Complex Tasks (e.g., ``I feel so hot, do you have any ideas?'') for IoT-enabled LLMs. Experimental validation across 22 sensor types and 6 microcontroller units demonstrates IoT-MCP's 100% task success rate to generate tool calls that fully meet expectations and obtain completely accurate results, 205ms average response time, and 74KB peak memory footprint. This work delivers both an open-source integration framework (https://github.com/Duke-CEI-Center/IoT-MCP-Servers) and a standardized evaluation methodology for LLM-IoT systems.

This paper introduces DOoM, a new open-source benchmark designed to assess the capabilities of language models in solving mathematics and physics problems in Russian. The benchmark includes problems of varying difficulty, ranging from school-level tasks to university Olympiad and entrance exam questions. In this paper we discuss the motivation behind its creation, describe dataset's structure and evaluation methodology, and present initial results from testing various models. Analysis of the results shows a correlation between model performance and the number of tokens used, and highlights differences in performance between mathematics and physics tasks.

Evaluating Computer Use Agents (CUAs) on interactive environments is fraught with methodological pitfalls that the field has yet to systematically address. We show that a 1MB replay script that blindly executes a recorded action sequence without ever observing the screen outperforms frontier models on prominent static benchmarks, and prove that its expected success rate is exactly equal to the source agent's pass@k in deterministic environments. We trace this and other failures to two root causes: non-principled environment design (static, unsandboxed, or unreliably verified environments) and non-principled evaluation methodology (naive aggregation and misuse of pass@k for stateful UI interactions). To address the first, we propose PRISM, five design principles for CUA environments (privileged verification, realistic environments, integrity-checked configurations, sandboxed execution, and multifactorial variability) and instantiate them in DigiWorld, a benchmark of 15 realistic sandboxed mobile applications able to evaluate agents in over 3.2 million verified unique configurations. To address the second, we develop an aggregation framework pairing Wilson score intervals with hierarchical bootstrap, producing confidence intervals that correctly account for the nested structure of CUA benchmarks, as we empirically demonstrate. All together, we show that principled environment design and rigorous evaluation methodology are not optional refinements but prerequisites for meaningful CUA research.

How much do large language models actually hallucinate when answering questions grounded in provided documents? Despite the critical importance of this question for enterprise AI deployments, reliable measurement has been hampered by benchmarks that rely on static datasets vulnerable to contamination, LLM-based judges with documented biases, or evaluation scales too small for statistical confidence. We address this gap using RIKER, a ground-truth-first evaluation methodology that enables deterministic scoring without human annotation. Across 35 open-weight models, three context lengths (32K, 128K, and 200K tokens), four temperature settings, and three hardware platforms (NVIDIA H200, AMD MI300X, and Intel Gaudi 3), we conducted over 172 billion tokens of evaluation - an order of magnitude beyond prior work. Our findings reveal that: (1) even the best-performing models fabricate answers at a non-trivial rate - 1.19% at best at 32K, with top-tier models at 5 - 7% - and fabrication rises steeply with context length, nearly tripling at 128K and exceeding 10% for all models at 200K; (2) model selection dominates all other factors, with overall accuracy spanning a 72-percentage-point range and model family predicting fabrication resistance better than model size; (3) temperature effects are nuanced - T=0.0 yields the best overall accuracy in roughly 60% of cases, but higher temperatures reduce fabrication for the majority of models and dramatically reduce coherence loss (infinite generation loops), which can reach 48x higher rates at T=0.0 versus T=1.0; (4) grounding ability and fabrication resistance are distinct capabilities - models that excel at finding facts may still fabricate facts that do not exist; and (5) results are consistent across hardware platforms, confirming that deployment decisions need not be hardware-dependent.

Embodied Artificial Intelligence (Embodied AI) is an emerging frontier in robotics, driven by the need for autonomous systems that can perceive, reason, and act in complex physical environments. While single-arm systems have shown strong task performance, collaborative dual-arm systems are essential for handling more intricate tasks involving rigid, deformable, and tactile-sensitive objects. To advance this goal, we launched the RoboTwin Dual-Arm Collaboration Challenge at the 2nd MEIS Workshop, CVPR 2025. Built on the RoboTwin Simulation platform (1.0 and 2.0) and the AgileX COBOT-Magic Robot platform, the competition consisted of three stages: Simulation Round 1, Simulation Round 2, and a final Real-World Round. Participants totally tackled 17 dual-arm manipulation tasks, covering rigid, deformable, and tactile-based scenarios. The challenge attracted 64 global teams and over 400 participants, producing top-performing solutions like SEM and AnchorDP3 and generating valuable insights into generalizable bimanual policy learning. This report outlines the competition setup, task design, evaluation methodology, key findings and future direction, aiming to support future research on robust and generalizable bimanual manipulation policies. The Challenge Webpage is available at https://robotwin-benchmark.github.io/cvpr-2025-challenge/.

Stealthy hardware Trojans (HTs) inserted during the fabrication of integrated circuits can bypass the security of critical infrastructures. Although researchers have proposed many techniques to detect HTs, several limitations exist, including: (i) a low success rate, (ii) high algorithmic complexity, and (iii) a large number of test patterns. Furthermore, the most pertinent drawback of prior detection techniques stems from an incorrect evaluation methodology, i.e., they assume that an adversary inserts HTs randomly. Such inappropriate adversarial assumptions enable detection techniques to claim high HT detection accuracy, leading to a "false sense of security." Unfortunately, to the best of our knowledge, despite more than a decade of research on detecting HTs inserted during fabrication, there have been no concerted efforts to perform a systematic evaluation of HT detection techniques. In this paper, we play the role of a realistic adversary and question the efficacy of HT detection techniques by developing an automated, scalable, and practical attack framework, ATTRITION, using reinforcement learning (RL). ATTRITION evades eight detection techniques across two HT detection categories, showcasing its agnostic behavior. ATTRITION achieves average attack success rates of 47times and 211times compared to randomly inserted HTs against state-of-the-art HT detection techniques. We demonstrate ATTRITION's ability to evade detection techniques by evaluating designs ranging from the widely-used academic suites to larger designs such as the open-source MIPS and mor1kx processors to AES and a GPS module. Additionally, we showcase the impact of ATTRITION-generated HTs through two case studies (privilege escalation and kill switch) on the mor1kx processor. We envision that our work, along with our released HT benchmarks and models, fosters the development of better HT detection techniques.

A MICRO 2024 best paper runner-up publication (the Mess paper) with all three artifact badges awarded (including "Reproducible") proposes a new benchmark to evaluate real and simulated memory system performance. In this paper, we demonstrate that the Ramulator 2.0 simulation results reported in the Mess paper are incorrect and, at the time of the publication of the Mess paper, irreproducible. We find that the authors of Mess paper made multiple trivial human errors in both the configuration and usage of the simulators. We show that by correctly configuring Ramulator 2.0, Ramulator 2.0's simulated memory system performance actually resembles real system characteristics well, and thus a key claimed contribution of the Mess paper is factually incorrect. We also identify that the DAMOV simulation results in the Mess paper use wrong simulation statistics that are unrelated to the simulated DRAM performance. Moreover, the Mess paper's artifact repository lacks the necessary sources to fully reproduce all the Mess paper's results. Our work corrects the Mess paper's errors regarding Ramulator 2.0 and identifies important issues in the Mess paper's memory simulator evaluation methodology. We emphasize the importance of both carefully and rigorously validating simulation results and contacting simulator authors and developers, in true open source spirit, to ensure these simulators are used with correct configurations and as intended. We encourage the computer architecture community to correct the Mess paper's errors. This is necessary to prevent the propagation of inaccurate and misleading results, and to maintain the reliability of the scientific record. Our investigation also opens up questions about the integrity of the review and artifact evaluation processes. To aid future work, our source code and scripts are openly available at https://github.com/CMU-SAFARI/ramulator2/tree/mess.

While multimodal conversation agents are gaining importance in several domains such as retail, travel etc., deep learning research in this area has been limited primarily due to the lack of availability of large-scale, open chatlogs. To overcome this bottleneck, in this paper we introduce the task of multimodal, domain-aware conversations, and propose the MMD benchmark dataset. This dataset was gathered by working in close coordination with large number of domain experts in the retail domain. These experts suggested various conversations flows and dialog states which are typically seen in multimodal conversations in the fashion domain. Keeping these flows and states in mind, we created a dataset consisting of over 150K conversation sessions between shoppers and sales agents, with the help of in-house annotators using a semi-automated manually intense iterative process. With this dataset, we propose 5 new sub-tasks for multimodal conversations along with their evaluation methodology. We also propose two multimodal neural models in the encode-attend-decode paradigm and demonstrate their performance on two of the sub-tasks, namely text response generation and best image response selection. These experiments serve to establish baseline performance and open new research directions for each of these sub-tasks. Further, for each of the sub-tasks, we present a `per-state evaluation' of 9 most significant dialog states, which would enable more focused research into understanding the challenges and complexities involved in each of these states.

Retrieval-augmented generation (RAG) is increasingly deployed in enterprise search and document-centric assistants, where responses must be grounded in long and complex source materials. In practice, verifying that generated answers faithfully reflect retrieved documents is difficult: large language models can check long contexts but are too slow and costly for interactive services, while lightweight classifiers operate within strict context limits and frequently miss evidence outside truncated passages. We present the design of a real-time verification component integrated into a production RAG pipeline that enables full-document grounding under latency constraints. The system processes documents up to 32K tokens and employs adaptive inference strategies to balance response time and verification coverage across workloads. We describe the architectural decisions, operational trade-offs, and evaluation methodology used to deploy the verifier, and show that full-context verification substantially improves detection of unsupported responses compared with truncated validation. Our experience highlights when long-context verification is necessary, why chunk-based checking often fails in real documents, and how latency budgets shape model design. These findings provide practical guidance for practitioners building reliable large-scale retrieval-augmented applications. (Model, benchmark, and code: https://huggingface.co/llm-semantic-router)

Multimodal large language models (MLLMs) have broadened the scope of AI applications. Existing automatic evaluation methodologies for MLLMs are mainly limited in evaluating queries without considering user experiences, inadequately addressing the nuances of creative and associative multimodal tasks. However, the open-ended and subjective nature of such tasks poses a significant challenge to the evaluation methodology, where it is difficult to define the ground-truth answers for them. To this end, in our paper, we propose a new evaluation paradigm for MLLMs, which is evaluating MLLMs with per-sample criteria using potent MLLM as the judge. To validate the feasibility and effectiveness of this paradigm, we design a benchmark, dubbed MLLM-Bench, by curating the evaluation samples across six comprehensive cognitive levels. We benchmark 21 popular MLLMs in a pairwise-comparison fashion, showing diverse performance across models. Moreover, the validity of our benchmark manifests itself in reaching 88.02% agreement with human evaluation. We contend that the proposed paradigm explores the potential of MLLMs as effective evaluation tools with the help of per-sample criteria. See online leaderboard at https://mllm-bench.llmzoo.com.

The Hitchhiker's Guide to Agentic AI is a comprehensive practitioner's reference for building autonomous AI systems. The book covers the full stack from first principles to production deployment, organized around a central thesis: building great agentic systems requires understanding every layer of the pipeline, not just one. The book opens with the LLM substrate -- transformer architecture, GPU systems, training and fine-tuning (SFT,LoRA, MoE), model compression, and inference optimization -- treated as essential foundations rather than the primary focus. It then develops the alignment and reasoning layer: reinforcement learning from human feedback (RLHF), PPO, DPO and its variants, GRPO, reward modeling, and RL for large reasoning models including chain-of-thought and test-time scaling. The second half is devoted to agentic AI proper. Topics include agentic training and trajectory-based RL, retrieval-augmented generation (RAG and Agentic RAG), memory systems (in-context, external, episodic, and semantic), agent harness design and context management, and a taxonomy of agent design patterns. Inter-agent coordination is covered in depth: the Model Context Protocol (MCP), agent skills and tool use, the Agent-to-Agent (A2A) communication protocol, and multi-agent architectures spanning centralized, decentralized, and hierarchical topologies. The book concludes with agent development frameworks, agentic UI design, evaluation methodology for agentic tasks, and production deployment. Each chapter pairs rigorous theoretical foundations with implementation guidance, code examples, and references to the primary literature.

Given the recent introduction of multiple language models and the ongoing demand for improved Natural Language Processing tasks, particularly summarization, this work provides a comprehensive benchmarking of 20 recent language models, focusing on smaller ones for the news summarization task. In this work, we systematically test the capabilities and effectiveness of these models in summarizing news article texts which are written in different styles and presented in three distinct datasets. Specifically, we focus in this study on zero-shot and few-shot learning settings and we apply a robust evaluation methodology that combines different evaluation concepts including automatic metrics, human evaluation, and LLM-as-a-judge. Interestingly, including demonstration examples in the few-shot learning setting did not enhance models' performance and, in some cases, even led to worse quality of the generated summaries. This issue arises mainly due to the poor quality of the gold summaries that have been used as reference summaries, which negatively impacts the models' performance. Furthermore, our study's results highlight the exceptional performance of GPT-3.5-Turbo and GPT-4, which generally dominate due to their advanced capabilities. However, among the public models evaluated, certain models such as Qwen1.5-7B, SOLAR-10.7B-Instruct-v1.0, Meta-Llama-3-8B and Zephyr-7B-Beta demonstrated promising results. These models showed significant potential, positioning them as competitive alternatives to large models for the task of news summarization.

Large language models (LLMs) are being rapidly integrated into decision-support tools, automation workflows, and AI-enabled software systems. However, their behavior in production environments remains poorly understood, and their failure patterns differ fundamentally from those of traditional machine learning models. This paper presents a system-level taxonomy of fifteen hidden failure modes that arise in real-world LLM applications, including multi-step reasoning drift, latent inconsistency, context-boundary degradation, incorrect tool invocation, version drift, and cost-driven performance collapse. Using this taxonomy, we analyze the growing gap in evaluation and monitoring practices: existing benchmarks measure knowledge or reasoning but provide little insight into stability, reproducibility, drift, or workflow integration. We further examine the production challenges associated with deploying LLMs - including observability limitations, cost constraints, and update-induced regressions - and outline high-level design principles for building reliable, maintainable, and cost-aware LLM systems. Finally, we outline high-level design principles for building reliable, maintainable, and cost-aware LLM-based systems. By framing LLM reliability as a system-engineering problem rather than a purely model-centric one, this work provides an analytical foundation for future research on evaluation methodology, AI system robustness, and dependable LLM deployment.

Academic posters are vital for scholarly communication, yet their manual creation is time-consuming. However, automated academic poster generation faces significant challenges in preserving intricate scientific details and achieving effective visual-textual integration. Existing approaches often struggle with semantic richness and structural nuances, and lack standardized benchmarks for evaluating generated academic posters comprehensively. To address these limitations, we introduce P2P, the first flexible, LLM-based multi-agent framework that generates high-quality, HTML-rendered academic posters directly from research papers, demonstrating strong potential for practical applications. P2P employs three specialized agents-for visual element processing, content generation, and final poster assembly-each integrated with dedicated checker modules to enable iterative refinement and ensure output quality. To foster advancements and rigorous evaluation in this domain, we construct and release P2PInstruct, the first large-scale instruction dataset comprising over 30,000 high-quality examples tailored for the academic paper-to-poster generation task. Furthermore, we establish P2PEval, a comprehensive benchmark featuring 121 paper-poster pairs and a dual evaluation methodology (Universal and Fine-Grained) that leverages LLM-as-a-Judge and detailed, human-annotated checklists. Our contributions aim to streamline research dissemination and provide the community with robust tools for developing and evaluating next-generation poster generation systems.

Knowledge of tree species distribution is fundamental to managing forests. New deep learning approaches promise significant accuracy gains for forest mapping, and are becoming a critical tool for mapping multiple tree species at scale. To advance the field, deep learning researchers need large benchmark datasets with high-quality annotations. To this end, we present the PureForest dataset: a large-scale, open, multimodal dataset designed for tree species classification from both Aerial Lidar Scanning (ALS) point clouds and Very High Resolution (VHR) aerial images. Most current public Lidar datasets for tree species classification have low diversity as they only span a small area of a few dozen annotated hectares at most. In contrast, PureForest has 18 tree species grouped into 13 semantic classes, and spans 339 km^2 across 449 distinct monospecific forests, and is to date the largest and most comprehensive Lidar dataset for the identification of tree species. By making PureForest publicly available, we hope to provide a challenging benchmark dataset to support the development of deep learning approaches for tree species identification from Lidar and/or aerial imagery. In this data paper, we describe the annotation workflow, the dataset, the recommended evaluation methodology, and establish a baseline performance from both 3D and 2D modalities.

Deep reinforcement learning (RL) algorithms are predominantly evaluated by comparing their relative performance on a large suite of tasks. Most published results on deep RL benchmarks compare point estimates of aggregate performance such as mean and median scores across tasks, ignoring the statistical uncertainty implied by the use of a finite number of training runs. Beginning with the Arcade Learning Environment (ALE), the shift towards computationally-demanding benchmarks has led to the practice of evaluating only a small number of runs per task, exacerbating the statistical uncertainty in point estimates. In this paper, we argue that reliable evaluation in the few run deep RL regime cannot ignore the uncertainty in results without running the risk of slowing down progress in the field. We illustrate this point using a case study on the Atari 100k benchmark, where we find substantial discrepancies between conclusions drawn from point estimates alone versus a more thorough statistical analysis. With the aim of increasing the field's confidence in reported results with a handful of runs, we advocate for reporting interval estimates of aggregate performance and propose performance profiles to account for the variability in results, as well as present more robust and efficient aggregate metrics, such as interquartile mean scores, to achieve small uncertainty in results. Using such statistical tools, we scrutinize performance evaluations of existing algorithms on other widely used RL benchmarks including the ALE, Procgen, and the DeepMind Control Suite, again revealing discrepancies in prior comparisons. Our findings call for a change in how we evaluate performance in deep RL, for which we present a more rigorous evaluation methodology, accompanied with an open-source library rliable, to prevent unreliable results from stagnating the field.

We present the 2017 DAVIS Challenge on Video Object Segmentation, a public dataset, benchmark, and competition specifically designed for the task of video object segmentation. Following the footsteps of other successful initiatives, such as ILSVRC and PASCAL VOC, which established the avenue of research in the fields of scene classification and semantic segmentation, the DAVIS Challenge comprises a dataset, an evaluation methodology, and a public competition with a dedicated workshop co-located with CVPR 2017. The DAVIS Challenge follows up on the recent publication of DAVIS (Densely-Annotated VIdeo Segmentation), which has fostered the development of several novel state-of-the-art video object segmentation techniques. In this paper we describe the scope of the benchmark, highlight the main characteristics of the dataset, define the evaluation metrics of the competition, and present a detailed analysis of the results of the participants to the challenge.

The use of NLP in the realm of financial technology is broad and complex, with applications ranging from sentiment analysis and named entity recognition to question answering. Large Language Models (LLMs) have been shown to be effective on a variety of tasks; however, no LLM specialized for the financial domain has been reported in literature. In this work, we present BloombergGPT, a 50 billion parameter language model that is trained on a wide range of financial data. We construct a 363 billion token dataset based on Bloomberg's extensive data sources, perhaps the largest domain-specific dataset yet, augmented with 345 billion tokens from general purpose datasets. We validate BloombergGPT on standard LLM benchmarks, open financial benchmarks, and a suite of internal benchmarks that most accurately reflect our intended usage. Our mixed dataset training leads to a model that outperforms existing models on financial tasks by significant margins without sacrificing performance on general LLM benchmarks. Additionally, we explain our modeling choices, training process, and evaluation methodology. As a next step, we plan to release training logs (Chronicles) detailing our experience in training BloombergGPT.

Predictive code completion greatly accelerates how quickly developers work. In spreadsheets, despite being much more common, such auto-completion features are virtually non-existent. To address this gap, we introduce a benchmark for systems that observe a sequence of user actions in a spreadsheet and predict future actions. Two challenges are (1) the absence of edit histories in public spreadsheet corpora and (2) the complex space of spreadsheet actions (spatial, temporal, composite). To address (1), we manually curate 52 sequences of 12K actions that recreate spreadsheets from public corpora, seeded by parametrized heuristics and LLM refinement. To address (2), we propose an online evaluation that expects a prediction after each user action, accepts or rejects that prediction, updates the future actions upon acceptance, and repeats this until the target spreadsheet is obtained. We use multiple baseline predictors (including zero-shot LLMs, fine-tuned SLMs, and classical models) and analyze different properties that our benchmark teaches us, including but not limited to: properties of saved actions and false positives, efficiency, effect of user profiles, effect of triggers, and effect of context.

While large language model (LLM)-based text-to-speech (TTS) systems have achieved high-quality speech synthesis, most existing systems focus on English and Chinese. Japanese, however, remains under-explored, and its unique linguistic challenges, such as widespread context-dependent kanji polyphony, have yet to be adequately tackled. Here we introduce Sarashina2.2-TTS (https://github.com/sbintuitions/sarashina2.2-tts), a Japanese-centric LLM-TTS system that tackles these challenges through a dual approach: data strategy and evaluation methodology. First, we scale training to approximately 361k hours of speech, incorporating a balanced mix of Japanese and English data. Furthermore, we design a targeted data augmentation pipeline covering all 2,136 Joyo (regular-use) kanji designated by Japan's Agency for Cultural Affairs to efficiently address kanji polyphony disambiguation. Second, we introduce the Joyo Kanji Yomi Benchmark (https://github.com/sbintuitions/JoyoKanji-Yomi-Benchmark), covering all 2,136 Joyo kanji and their 4,378 readings. Alongside this benchmark, we propose Kana-CER, a metric that compares synthesized speech against reference readings in the kana space, eliminating orthographic variations to directly measure pronunciation correctness. Experiments demonstrate that our targeted data augmentation significantly improves reading accuracy. Overall, Sarashina2.2-TTS achieves state-of-the-art kanji-level reading accuracy and matches top baselines on general sentence-level pronunciation, while delivering the highest speaker similarity in zero-shot Japanese speech synthesis. Furthermore, cross-lingual evaluation reveals that Sarashina2.2-TTS is the only system that maintains stable Japanese pronunciation regardless of the prompt language, confirming that our balanced training approach improves cross-lingual robustness.

We conduct a systematic audit of three widely used reasoning benchmarks, SocialIQa, FauxPas-EAI, and ToMi, and uncover pervasive flaws in both benchmark items and evaluation methodology. Using five LLMs (GPT-{3, 3.5, 4, o1}, and LLaMA 3.1) as diagnostic tools, we identify structural, semantic, and pragmatic issues in benchmark design (e.g., duplicated items, ambiguous wording, and implausible answers), as well as scoring procedures that prioritize output form over reasoning process. Through systematic human annotation and re-evaluation on cleaned benchmark subsets, we find that model scores often improve not due to due to erratic surface wording variations and not to improved reasoning. Infact, further analyses show that model performance is highly sensitive to minor input variations such as context availability and phrasing, revealing that high scores may reflect alignment with format-specific cues rather than consistent inference based on the input. These findings challenge the validity of current benchmark-based claims about reasoning in LLMs, and highlight the need for evaluation protocols that assess reasoning as a process of drawing inference from available information, rather than as static output selection. We release audited data and evaluation tools to support more interpretable and diagnostic assessments of model reasoning.

Large-scale Language Models (LLMs) have revolutionized human-AI interaction and achieved significant success in the generation of novel ideas. However, current assessments of idea generation overlook crucial factors such as knowledge leakage in LLMs, the absence of open-ended benchmarks with grounded truth, and the limited scope of feasibility analysis constrained by prompt design. These limitations hinder the potential of uncovering groundbreaking research ideas. In this paper, we present AI Idea Bench 2025, a framework designed to quantitatively evaluate and compare the ideas generated by LLMs within the domain of AI research from diverse perspectives. The framework comprises a comprehensive dataset of 3,495 AI papers and their associated inspired works, along with a robust evaluation methodology. This evaluation system gauges idea quality in two dimensions: alignment with the ground-truth content of the original papers and judgment based on general reference material. AI Idea Bench 2025's benchmarking system stands to be an invaluable resource for assessing and comparing idea-generation techniques, thereby facilitating the automation of scientific discovery.

Pretrained language models (PLMs) display impressive performances and have captured the attention of the NLP community. Establishing best practices in pretraining has, therefore, become a major focus of NLP research, especially since insights gained from monolingual English models may not necessarily apply to more complex multilingual models. One significant caveat of the current state of the art is that different works are rarely comparable: they often discuss different parameter counts, training data, and evaluation methodology. This paper proposes a comparison of multilingual pretraining objectives in a controlled methodological environment. We ensure that training data and model architectures are comparable, and discuss the downstream performances across 6 languages that we observe in probing and fine-tuning scenarios. We make two key observations: (1) the architecture dictates which pretraining objective is optimal; (2) multilingual translation is a very effective pretraining objective under the right conditions. We make our code, data, and model weights available at \url{https://github.com/Helsinki-NLP/lm-vs-mt}.

The Model Context Protocol (MCP) is rapidly emerging as a pivotal open standard, designed to enhance agent-tool integration and interoperability, and is positioned to unlock a new era of powerful, interconnected, and genuinely utilitarian agentic AI. However, despite MCP's growing adoption, existing benchmarks often fail to capture real-world agent performance within this new paradigm, leading to a distorted perception of their true operational value and an inability to reliably differentiate proficiencies. To bridge this critical evaluation gap, we introduce MCP-AgentBench -- a comprehensive benchmark specifically engineered to rigorously assess language agent capabilities in MCP-mediated tool interactions. Core contributions of MCP-AgentBench include: the establishment of a robust MCP testbed comprising 33 operational servers with 188 distinct tools; the development of a benchmark featuring 600 systematically designed queries distributed across 6 distinct categories of varying interaction complexity; and the introduction of MCP-Eval, a novel outcome-oriented evaluation methodology prioritizing real-world task success. Through extensive empirical evaluation of leading language agents, we provide foundational insights. MCP-AgentBench aims to equip the research community with a standardized and reliable framework to build, validate, and advance agents capable of fully leveraging MCP's transformative benefits, thereby accelerating progress toward truly capable and interoperable AI systems.

Models like OpenAI-o3 pioneer visual grounded reasoning by dynamically referencing visual regions, just like human "thinking with images". However, no benchmark exists to evaluate these capabilities holistically. To bridge this gap, we propose TreeBench (Traceable Evidence Evaluation Benchmark), a diagnostic benchmark built on three principles: (1) focused visual perception of subtle targets in complex scenes, (2) traceable evidence via bounding box evaluation, and (3) second-order reasoning to test object interactions and spatial hierarchies beyond simple object localization. Prioritizing images with dense objects, we initially sample 1K high-quality images from SA-1B, and incorporate eight LMM experts to manually annotate questions, candidate options, and answers for each image. After three stages of quality control, TreeBench consists of 405 challenging visual question-answering pairs, even the most advanced models struggle with this benchmark, where none of them reach 60% accuracy, e.g., OpenAI-o3 scores only 54.87. Furthermore, we introduce TreeVGR (Traceable Evidence Enhanced Visual Grounded Reasoning), a training paradigm to supervise localization and reasoning jointly with reinforcement learning, enabling accurate localizations and explainable reasoning pathways. Initialized from Qwen2.5-VL-7B, it improves V* Bench (+16.8), MME-RealWorld (+12.6), and TreeBench (+13.4), proving traceability is key to advancing vision-grounded reasoning. The code is available at https://github.com/Haochen-Wang409/TreeVGR.

The computer vision and robotics research communities are each strong. However progress in computer vision has become turbo-charged in recent years due to big data, GPU computing, novel learning algorithms and a very effective research methodology. By comparison, progress in robotics seems slower. It is true that robotics came later to exploring the potential of learning -- the advantages over the well-established body of knowledge in dynamics, kinematics, planning and control is still being debated, although reinforcement learning seems to offer real potential. However, the rapid development of computer vision compared to robotics cannot be only attributed to the former's adoption of deep learning. In this paper, we argue that the gains in computer vision are due to research methodology -- evaluation under strict constraints versus experiments; bold numbers versus videos.

This paper presents a comprehensive collection of industrial screw driving datasets designed to advance research in manufacturing process monitoring and quality control. The collection comprises six distinct datasets with over 34,000 individual screw driving operations conducted under controlled experimental conditions, capturing the multifaceted nature of screw driving processes in plastic components. Each dataset systematically investigates specific aspects: natural thread degradation patterns through repeated use (s01), variations in surface friction conditions including contamination and surface treatments (s02), diverse assembly faults with up to 27 error types (s03-s04), and fabrication parameter variations in both upper and lower workpieces through modified injection molding settings (s05-s06). We detail the standardized experimental setup used across all datasets, including hardware specifications, process phases, and data acquisition methods. The hierarchical data model preserves the temporal and operational structure of screw driving processes, facilitating both exploratory analysis and the development of machine learning models. To maximize accessibility, we provide dual access pathways: raw data through Zenodo with a persistent DOI, and a purpose-built Python library (PyScrew) that offers consistent interfaces for data loading, preprocessing, and integration with common analysis workflows. These datasets serve diverse research applications including anomaly detection, predictive maintenance, quality control system development, feature extraction methodology evaluation, and classification of specific error conditions. By addressing the scarcity of standardized, comprehensive datasets in industrial manufacturing, this collection enables reproducible research and fair comparison of analytical approaches in an area of growing importance for industrial automation.

This technical report presents the training methodology and evaluation results of the open-source multilingual E5 text embedding models, released in mid-2023. Three embedding models of different sizes (small / base / large) are provided, offering a balance between the inference efficiency and embedding quality. The training procedure adheres to the English E5 model recipe, involving contrastive pre-training on 1 billion multilingual text pairs, followed by fine-tuning on a combination of labeled datasets. Additionally, we introduce a new instruction-tuned embedding model, whose performance is on par with state-of-the-art, English-only models of similar sizes. Information regarding the model release can be found at https://github.com/microsoft/unilm/tree/master/e5 .

This technical report presents the training methodology and evaluation results of the open-source Jasper-Token-Compression-600M model, released in November 2025. Building on previous distillation-based recipes from the English Stella and Jasper models, we successfully extend this approach to a bilingual (English and Chinese) domain, further enhancing model performance through the incorporation of contrastive learning. A key innovation of our model is the introduction of a one-dimensional convolution-based token compression module. We dynamically adjust the compression rate during training, enabling the model to learn more robust and efficient compressed text representations. By combining knowledge distillation with token compression techniques, we achieve significant improvements in both embedding quality and inference efficiency. Our model performs with higher efficiency than a traditional 0.6B model while achieving performance comparable to that of an 8B model. For more information on the model release, visit: https://huggingface.co/infgrad/Jasper-Token-Compression-600M.

This technical report presents the training methodology and evaluation results of the open-source dewey_en_beta embedding model. The increasing demand for retrieval-augmented generation (RAG) systems and the expanding context window capabilities of large language models (LLMs) have created critical challenges for conventional embedding models. Current approaches often struggle to maintain semantic coherence when processing documents exceeding typical sequence length limitations, significantly impacting retrieval performance in knowledge-intensive applications. This paper presents dewey_en_beta, a novel text embedding model that achieves excellent performance on MTEB (Eng, v2) and LongEmbed benchmark while supporting 128K token sequences. Our technical contribution centers on chunk alignment training, an innovative methodology that enables the simultaneous generation of localized chunk embeddings and global document-level representations through distillation. Information regarding the model release can be found at https://huggingface.co/infgrad/dewey_en_beta.

This paper poses two critical issues in evaluating base models (without post-training): (1) Unstable evaluation during training: in the early stages of pre-training, the models lack the capability to answer questions as required, leading to unstable evaluation results. This instability makes it difficult to provide solid conclusions to guide the training, especially for key experiments such as data ablation and scaling law. (2) Inconsistency between base and instruct models: base models generally exhibit poorer evaluation performance compared to corresponding instruct models. This gap poses a challenge for assessing whether a base model with better evaluation can truly lead to a better instruct model. To address these issues, we propose Base model Oriented Systematic Evaluation (BOSE), a method specifically designed to optimize the evaluation of base models. Specifically, BOSE introduces two key innovations: In-Context Light-instruction Prompt (ICLiP) for open-ended tasks and Blank-ppl for multi-choice tasks with candidate options, which transforms the standard perplexity (ppl) metric into a fill-in-the-blank format to mitigate early-stage evaluation fluctuations. Furthermore, we are the first to propose Kendall's rank correlation to quantitatively measure the evaluation stability and consistency. Experimental results demonstrate that BOSE significantly enhances both the stability of evaluations during pre-training and the consistency between base and instruct models, thereby providing more reliable guidance for the LLMs' training.

As large language models (LLMs) become increasingly common in educational applications, there is a growing need for evidence-based methods to design and evaluate LLM prompts that produce personalized and pedagogically aligned out-puts. This study presents a generalizable, systematic approach for evaluating prompts, demonstrated through an analysis of LLM-generated follow-up questions in a structured dialogue activity. Six prompt templates were designed and tested. The templates incorporated established prompt engineering patterns, with each prompt emphasizing distinct pedagogical strategies. The prompt templates were compared through a tournament-style evaluation framework that can be adapted for other educational applications. The tournament employed the Glicko2 rating system with eight judges evaluating question pairs across three dimensions: format, dialogue support, and appropriateness for learners. Data was sourced from 120 authentic user interactions across three distinct educational deployments. Results showed that a single prompt related to strategic reading out-performed other templates with win probabilities ranging from 81% to 100% in pairwise comparisons. This prompt combined persona and context manager pat-terns and was designed to support metacognitive learning strategies such as self-directed learning. The methodology showcases how educational technology re- searchers can systematically evaluate and improve prompt designs, moving beyond ad-hoc prompt engineering toward evidence-based prompt development for educational applications.

The rapid rollout of AI in heterogeneous public and societal sectors has subsequently escalated the need for compliance with regulatory standards and frameworks. The EU AI Act has emerged as a landmark in the regulatory landscape. The development of solutions that elicit the level of AI systems' compliance with such standards is often limited by the lack of resources, hindering the semi-automated or automated evaluation of their performance. This generates the need for manual work, which is often error-prone, resource-limited or limited to cases not clearly described by the regulation. This paper presents an open, transparent, and reproducible method of creating a resource that facilitates the evaluation of NLP models with a strong focus on RAG systems. We have developed a dataset that contain the tasks of risk-level classification, article retrieval, obligation generation, and question-answering for the EU AI Act. The dataset files are in a machine-to-machine appropriate format. To generate the files, we utilise domain knowledge as an exegetical basis, combining with the processing and reasoning power of large language models to generate scenarios along with the respective tasks. Our methodology demonstrates a way to harness language models for grounded generation with high document relevancy. Besides, we overcome limitations such as navigating the decision boundaries of risk-levels that are not explicitly defined within the EU AI Act, such as limited and minimal cases. Finally, we demonstrate our dataset's effectiveness by evaluating a RAG-based solution that reaches 0.87 and 0.85 F1-score for prohibited and high-risk scenarios.

As machine learning models evolve, maintaining transparency demands more human-centric explainable AI techniques. Counterfactual explanations, with roots in human reasoning, identify the minimal input changes needed to obtain a given output and, hence, are crucial for supporting decision-making. Despite their importance, the evaluation of these explanations often lacks grounding in user studies and remains fragmented, with existing metrics not fully capturing human perspectives. To address this challenge, we developed a diverse set of 30 counterfactual scenarios and collected ratings across 8 evaluation metrics from 206 respondents. Subsequently, we fine-tuned different Large Language Models (LLMs) to predict average or individual human judgment across these metrics. Our methodology allowed LLMs to achieve an accuracy of up to 63% in zero-shot evaluations and 85% (over a 3-classes prediction) with fine-tuning across all metrics. The fine-tuned models predicting human ratings offer better comparability and scalability in evaluating different counterfactual explanation frameworks.

We present methodology for using dynamic evaluation to improve neural sequence models. Models are adapted to recent history via a gradient descent based mechanism, causing them to assign higher probabilities to re-occurring sequential patterns. Dynamic evaluation outperforms existing adaptation approaches in our comparisons. Dynamic evaluation improves the state-of-the-art word-level perplexities on the Penn Treebank and WikiText-2 datasets to 51.1 and 44.3 respectively, and the state-of-the-art character-level cross-entropies on the text8 and Hutter Prize datasets to 1.19 bits/char and 1.08 bits/char respectively.

Recent advances in large language models (LLMs) for mathematical reasoning have largely focused on tasks with easily verifiable final answers; however, generating and verifying natural language math proofs remains an open challenge. We identify the absence of a reliable, fine-grained evaluator for LLM-generated math proofs as a critical gap. To address this, we propose a systematic methodology for developing and validating evaluators that assign fine-grained scores on a 0-7 scale to model-generated math proofs. To enable this study, we introduce ProofBench, the first expert-annotated dataset of fine-grained proof ratings, spanning 145 problems from six major math competitions (USAMO, IMO, Putnam, etc) and 435 LLM-generated solutions from Gemini-2.5-pro, o3, and DeepSeek-R1. %with expert gradings. Using ProofBench as a testbed, we systematically explore the evaluator design space across key axes: the backbone model, input context, instructions and evaluation workflow. Our analysis delivers ProofGrader, an evaluator that combines a strong reasoning backbone LM, rich context from reference solutions and marking schemes, and a simple ensembling method; it achieves a low Mean Absolute Error (MAE) of 0.926 against expert scores, significantly outperforming naive baselines. Finally, we demonstrate its practical utility in a best-of-n selection task: at n=16, ProofGrader achieves an average score of 4.14 (out of 7), closing 78% of the gap between a naive binary evaluator (2.48) and the human oracle (4.62), highlighting its potential to advance downstream proof generation.

Limited large-scale evaluations exist for facilitation strategies of online discussions due to significant costs associated with human involvement. An effective solution is synthetic discussion simulations using Large Language Models (LLMs) to create initial pilot experiments. We propose a simple, generalizable, LLM-driven methodology to prototype the development of LLM facilitators, and produce high-quality synthetic data without human involvement. We use our methodology to test whether current facilitation strategies can improve the performance of LLM facilitators. We find that, while LLM facilitators significantly improve synthetic discussions, there is no evidence that the application of more elaborate facilitation strategies proposed in modern Social Science research lead to further improvements in discussion quality, compared to more basic approaches. Additionally, we find that small LLMs (such as Mistral Nemo 12B) can perform comparably to larger models (such as LLaMa 70B), and that special instructions must be used for instruction-tuned models to induce toxicity in synthetic discussions. We confirm that each component of our methodology contributes substantially to high quality data via an ablation study. We release an open-source framework, "SynDisco" (pip install syndisco), which implements our methodology. We also release the "Virtual Moderation Dataset" (https://paperswithcode.com/dataset/vmd), a large, publicly available dataset containing LLM-generated and LLM-annotated discussions using multiple open-source LLMs.

Large language model unlearning aims to remove harmful information that LLMs have learnt to prevent their use for malicious purposes. LLMU and RMU have been proposed as two methods for LLM unlearning, achieving impressive results on unlearning benchmarks. We study in detail the impact of unlearning on LLM performance metrics using the WMDP dataset as well as a new biology dataset we create. We show that unlearning has a notable impact on general model capabilities, with the performance degradation being more significant in general for LLMU. We further test the robustness of the two methods and find that doing 5-shot prompting or rephrasing the question in simple ways can lead to an over ten-fold increase in accuracy on unlearning benchmarks. Finally, we show that training on unrelated data can almost completely recover pre-unlearning performance, demonstrating that these methods fail at truly unlearning. Our methodology serves as an evaluation framework for LLM unlearning methods. The code is available at: https://github.com/JaiDoshi/Knowledge-Erasure.

Factual questions typically can be answered correctly at different levels of granularity. For example, both ``August 4, 1961'' and ``1961'' are correct answers to the question ``When was Barack Obama born?''. Standard question answering (QA) evaluation protocols, however, do not explicitly take this into account and compare a predicted answer against answers of a single granularity level. In this work, we propose GRANOLA QA, a novel evaluation setting where a predicted answer is evaluated in terms of accuracy and informativeness against a set of multi-granularity answers. We present a simple methodology for enriching existing datasets with multi-granularity answers, and create GRANOLA-EQ, a multi-granularity version of the EntityQuestions dataset. We evaluate a range of decoding methods on GRANOLA-EQ, including a new algorithm, called Decoding with Response Aggregation (DRAG), that is geared towards aligning the response granularity with the model's uncertainty. Our experiments show that large language models with standard decoding tend to generate specific answers, which are often incorrect. In contrast, when evaluated on multi-granularity answers, DRAG yields a nearly 20 point increase in accuracy on average, which further increases for rare entities. Overall, this reveals that standard evaluation and decoding schemes may significantly underestimate the knowledge encapsulated in LMs.

Ensuring safe and effective use of AI requires understanding and anticipating its performance on novel tasks, from advanced scientific challenges to transformed workplace activities. So far, benchmarking has guided progress in AI, but it has offered limited explanatory and predictive power for general-purpose AI systems, given the low transferability across diverse tasks. In this paper, we introduce general scales for AI evaluation that can explain what common AI benchmarks really measure, extract ability profiles of AI systems, and predict their performance for new task instances, in- and out-of-distribution. Our fully-automated methodology builds on 18 newly-crafted rubrics that place instance demands on general scales that do not saturate. Illustrated for 15 large language models and 63 tasks, high explanatory power is unleashed from inspecting the demand and ability profiles, bringing insights on the sensitivity and specificity exhibited by different benchmarks, and how knowledge, metacognition and reasoning are affected by model size, chain-of-thought and distillation. Surprisingly, high predictive power at the instance level becomes possible using these demand levels, providing superior estimates over black-box baseline predictors based on embeddings or finetuning, especially in out-of-distribution settings (new tasks and new benchmarks). The scales, rubrics, battery, techniques and results presented here represent a major step for AI evaluation, underpinning the reliable deployment of AI in the years ahead. (Collaborative platform: https://kinds-of-intelligence-cfi.github.io/ADELE.)

Laboratory workflows in pharmaceutical and biomedical research encode substantial tacit knowledge -- expert judgment about failure conditions, decision branching logic, and contextual dependencies -- that remains inaccessible to protocol documents, sensor streams, and existing biomedical ontologies. We present a repeatable structured expert elicitation methodology and federated Semantic Knowledge Graph (SKG) architecture for capturing and querying this knowledge, demonstrated through deployment at the Biochemical and Cellular Pharmacology Department of Genentech. Knowledge is elicited via the Protocol Intelligence Co-pilot, a purpose-built AI interview agent that applies structured elicitation lenses to surface tacit procedural knowledge with expert-assigned confidence scores, producing graph representations across three tiers: program-level decision milestones, assay protocol knowledge, and physical execution infrastructure. Separately constructed subgraphs, exemplified by immunoassay (ELISA), quantitative mass spectrometry (LC-MS/PRM), and laboratory automation, are aligned through a shared upper ontology and queried as a single federated graph. Evaluation demonstrates seven query types structurally unavailable from any individual data source, including a cross-subgraph traversal that identifies automation-masked silent failures -- conditions where execution logs report success while scientific validity is compromised. Critically, the MASKED_BY graph relationship encodes a class of laboratory risk invisible to current informatics platforms -- the structural gap that prevents existing systems from reasoning about scientific validity. This architecture provides the semantic world model that AI laboratory agents currently lack: a queryable representation of where workflows fail silently, where human judgment is irreplaceable, and which execution assets mask rather than detect failure.

With the widespread adoption of large language models (LLMs) in code generation tasks, geospatial code generation has emerged as a critical frontier in the integration of artificial intelligence and geoscientific analysis. This trend underscores the urgent need for systematic evaluation methodologies to assess LLMs generation capabilities in geospatial contexts. In particular, geospatial computation and visualization tasks in JavaScript environments rely heavily on orchestrating diverse frontend libraries and ecosystems, placing elevated demands on a model's semantic understanding and code synthesis abilities. To address this challenge, we propose GeoJSEval--the first multimodal, function-level automatic evaluation framework for LLMs in JavaScript-based geospatial code generation. GeoJSEval comprises three core components: a standardized test suite (GeoJSEval-Bench), a code submission engine, and an evaluation module. It includes 432 function-level tasks and 2,071 structured test cases spanning five widely used JavaScript geospatial libraries and 25 mainstream geospatial data types. GeoJSEval enables multidimensional quantitative evaluation across metrics such as accuracy, output stability, execution efficiency, resource consumption, and error type distribution, and integrates boundary testing mechanisms to enhance robustness and coverage. We conduct a comprehensive evaluation of 18 state-of-the-art LLMs using GeoJSEval, revealing significant performance disparities and bottlenecks in spatial semantic understanding, code reliability, and function invocation accuracy. GeoJSEval provides a foundational methodology, evaluation resource, and practical toolkit for the standardized assessment and optimization of geospatial code generation models, with strong extensibility and applicability in real-world scenarios.

Evaluating robotics policies across thousands of environments and thousands of tasks is infeasible with existing approaches. This motivates the need for a new methodology for scalable robotics policy evaluation. In this paper, we propose dWorldEval, which uses a discrete diffusion world model as a scalable evaluation proxy for robotics policies. Specifically, dWorldEval maps all modalities - including vision, language, and robotic actions - into a unified token space, modeling them via a single transformer-based denoising network. In this paper, we propose dWorldEval, using a discrete diffusion world model as a scalable evaluation proxy for robotics policy. Specifically, it maps all modalities, including vision, language, and robotics action into a unified token space, then denoises them with a single transformer network. Building on this architecture, we employ a sparse keyframe memory to maintain spatiotemporal consistency. We also introduce a progress token that indicates the degree of task completion. At inference, the model jointly predicts future observations and progress token, allowing automatically determine success when the progress reaches 1. Extensive experiments demonstrate that dWorldEval significantly outperforms previous approaches, i.e., WorldEval, Ctrl-World, and WorldGym, on LIBERO, RoboTwin, and multiple real-robot tasks. It paves the way for a new architectural paradigm in building world simulators for robotics evaluation at scale.

As part of the Open Language Data Initiative shared tasks, we have expanded the FLORES+ evaluation set to include Emakhuwa, a low-resource language widely spoken in Mozambique. We translated the dev and devtest sets from Portuguese into Emakhuwa, and we detail the translation process and quality assurance measures used. Our methodology involved various quality checks, including post-editing and adequacy assessments. The resulting datasets consist of multiple reference sentences for each source. We present baseline results from training a Neural Machine Translation system and fine-tuning existing multilingual translation models. Our findings suggest that spelling inconsistencies remain a challenge in Emakhuwa. Additionally, the baseline models underperformed on this evaluation set, underscoring the necessity for further research to enhance machine translation quality for Emakhuwa. The data is publicly available at https://huggingface.co/datasets/LIACC/Emakhuwa-FLORES.

The rapid advancement of Large Language Models (LLMs) has driven their expanding application across various fields. One of the most promising applications is their role as evaluators based on natural language responses, referred to as ''LLMs-as-judges''. This framework has attracted growing attention from both academia and industry due to their excellent effectiveness, ability to generalize across tasks, and interpretability in the form of natural language. This paper presents a comprehensive survey of the LLMs-as-judges paradigm from five key perspectives: Functionality, Methodology, Applications, Meta-evaluation, and Limitations. We begin by providing a systematic definition of LLMs-as-Judges and introduce their functionality (Why use LLM judges?). Then we address methodology to construct an evaluation system with LLMs (How to use LLM judges?). Additionally, we investigate the potential domains for their application (Where to use LLM judges?) and discuss methods for evaluating them in various contexts (How to evaluate LLM judges?). Finally, we provide a detailed analysis of the limitations of LLM judges and discuss potential future directions. Through a structured and comprehensive analysis, we aim aims to provide insights on the development and application of LLMs-as-judges in both research and practice. We will continue to maintain the relevant resource list at https://github.com/CSHaitao/Awesome-LLMs-as-Judges.

Software architecture research relies on key architectural artifacts -- Software Architectures, Reference Architectures, and Architectural Frameworks -- that underpin the design and analysis of complex systems. Evaluating these artifacts is essential to assess tradeoffs and risks affecting quality attributes such as performance, modifiability, and security. Although methodologies like the Architecture Tradeoff Analysis Method (ATAM) support software architecture evaluation, their industrial focus misaligns with the IMRaD (Introduction, Methods, Results, Discussion) format prevalent in academic research, impeding transparency and reproducibility. Our prior work introduced the Architecture Tradeoff and Risk Analysis Framework (ATRAF), extending ATAM through three methods -- ATRAM, RATRAM, and AFTRAM, addressing all abstraction levels, using a unified, iterative four-phase spiral model. These phases -- Scenario and Requirements Gathering, Architectural Views and Scenario Realization, Attribute-Specific Analyses, and Sensitivity, Tradeoff, and Risk Analysis -- ensure traceability and coherence. This paper presents the ATRAF-driven IMRaD Methodology, a concise method to align ATRAF's phases with IMRaD sections. This methodology enhances the rigor, transparency, and accessibility of software architecture research, enabling systematic reporting of complex evaluations.

Matbench Discovery simulates the deployment of machine learning (ML) energy models in a high-throughput search for stable inorganic crystals. We address the disconnect between (i) thermodynamic stability and formation energy and (ii) in-domain vs out-of-distribution performance. Alongside this paper, we publish a Python package to aid with future model submissions and a growing online leaderboard with further insights into trade-offs between various performance metrics. To answer the question which ML methodology performs best at materials discovery, our initial release explores a variety of models including random forests, graph neural networks (GNN), one-shot predictors, iterative Bayesian optimizers and universal interatomic potentials (UIP). Ranked best-to-worst by their test set F1 score on thermodynamic stability prediction, we find CHGNet > M3GNet > MACE > ALIGNN > MEGNet > CGCNN > CGCNN+P > Wrenformer > BOWSR > Voronoi tessellation fingerprints with random forest. The top 3 models are UIPs, the winning methodology for ML-guided materials discovery, achieving F1 scores of ~0.6 for crystal stability classification and discovery acceleration factors (DAF) of up to 5x on the first 10k most stable predictions compared to dummy selection from our test set. We also highlight a sharp disconnect between commonly used global regression metrics and more task-relevant classification metrics. Accurate regressors are susceptible to unexpectedly high false-positive rates if those accurate predictions lie close to the decision boundary at 0 eV/atom above the convex hull where most materials are. Our results highlight the need to focus on classification metrics that actually correlate with improved stability hit rate.

"Citizen queries" are questions asked by an individual about government policies, guidance, and services that are relevant to their circumstances, encompassing a range of topics including benefits, taxes, immigration, employment, public health, and more. This represents a compelling use case for Large Language Models (LLMs) that respond to citizen queries with information that is adapted to a user's context and communicated according to their needs. However, in this use case, any misinformation could have severe, negative, likely invisible ramifications for an individual placing their trust in a model's response. To this effect, we introduce CitizenQuery-UK, a benchmark dataset of 22 thousand pairs of citizen queries and responses that have been synthetically generated from the swathes of public information on gov.uk about government in the UK. We present the curation methodology behind CitizenQuery-UK and an overview of its contents. We also introduce a methodology for the benchmarking of LLMs with the dataset, using an adaptation of FActScore to benchmark 11 models for factuality, abstention frequency, and verbosity. We document these results, and interpret them in the context of the public sector, finding that: (i) there are distinct performance profiles across model families, but each is competitive; (ii) high variance undermines utility; (iii) abstention is low and verbosity is high, with implications on reliability; and (iv) more trustworthy AI requires acknowledged "fallibility" in the way it interacts with users. The contribution of our research lies in assessing the trustworthiness of LLMs in citizen query tasks; as we see a world of increasing AI integration into day-to-day life, our benchmark, built entirely on open data, lays the foundations for better evidenced decision-making regarding AI and the public sector.

Large language models (LLMs) have demonstrated excellent capabilities in generating structured diagrams from natural language descriptions. In particular, they have shown great promise in generating sequence diagrams for software engineering, typically represented in a text-based syntax such as Mermaid. However, systematic evaluations in this space remain underdeveloped as there is a lack of existing benchmarks to assess the LLM's correctness in this task. To address this shortcoming, we introduce MermaidSeqBench, a human-verified and LLM-synthetically-extended benchmark for assessing an LLM's capabilities in generating Mermaid sequence diagrams from textual prompts. The benchmark consists of a core set of 132 samples, starting from a small set of manually crafted and verified flows. These were expanded via a hybrid methodology combining human annotation, in-context LLM prompting, and rule-based variation generation. Our benchmark uses an LLM-as-a-judge model to assess Mermaid sequence diagram generation across fine-grained metrics, including syntax correctness, activation handling, error handling, and practical usability. We perform initial evaluations on numerous state-of-the-art LLMs and utilize multiple LLM judge models to demonstrate the effectiveness and flexibility of our benchmark. Our results reveal significant capability gaps across models and evaluation modes. Our proposed benchmark provides a foundation for advancing research in structured diagram generation and for developing more rigorous, fine-grained evaluation methodologies.

A recent focus of large language model (LLM) development, as exemplified by generative search engines, is to incorporate external references to generate and support their claims. However, evaluating the attribution, i.e., verifying whether the generated statement is indeed fully supported by the cited reference, remains an open problem. Although human evaluation is common practice, it is costly and time-consuming. In this paper, we investigate the automatic evaluation of attribution by LLMs. We begin by providing a definition of attribution and then explore two approaches for automatic evaluation: prompting LLMs and fine-tuning smaller LMs. The fine-tuning data is repurposed from related tasks, such as question answering, fact-checking, natural language inference, and summarization. To facilitate the evaluation, we manually curate a set of test examples covering 12 domains from a generative search engine, New Bing. Our results on the curated test set and simulated test examples from existing benchmark questions highlight both promising signals as well as remaining challenges for the automatic evaluation of attribution. We hope our testbed, modeling methodology, and insights will help lay the foundation for future studies on this important problem.