Daily Papers

Common evaluation paradigms for language models focus on scoring single responses through accuracy metrics or proper scoring rules, failing to capture the full richness of a model's belief state. Recent work illustrates that language models hallucinate in-part because they are optimised to be good test-takers under binary scoring schemes that reward any answer over abstention. While this insight naturally leads to penalty-based approaches, they ignore crucial distinctions in how models distribute uncertainty, for example between hedging toward incorrect answers versus hedging toward "I don't know" responses. A novel evaluation metric, the Distributional Correctness Score (DCS), is introduced to solve this problem, i.e., of not considering a model's entire probability distribution over answer choices. DCS naturally distinguishes between harmful overconfidence in wrong answers and uncertainty expressed through abstention, providing scores in an interpretable default range. Through theoretical analysis and illustrative examples, DCS is demonstrated to offer a more nuanced and aligned evaluation paradigm that incentivises models to express genuine uncertainty rather than guessing. Adapting 12 existing evaluation benchmarks to DCS's variants and measuring performance on six language models reveals that for half of the tested benchmarks scores are negative across all tested models, indicating significant tendencies towards hallucination.

Retrieval-Augmented Generation (RAG) enriches Large Language Models (LLMs) by combining their internal, parametric knowledge with external, non-parametric sources, with the goal of improving factual correctness and minimizing hallucinations. The LiveRAG 2025 challenge explores RAG solutions to maximize accuracy on DataMorgana's QA pairs, which are composed of single-hop and multi-hop questions. The challenge provides access to sparse OpenSearch and dense Pinecone indices of the Fineweb 10BT dataset. It restricts model use to LLMs with up to 10B parameters and final answer generation with Falcon-3-10B. A judge-LLM assesses the submitted answers along with human evaluators. By exploring distinct retriever combinations and RAG solutions under the challenge conditions, our final solution emerged using InstructRAG in combination with a Pinecone retriever and a BGE reranker. Our solution achieved a correctness score of 1.13 and a faithfulness score of 0.55, placing fourth in the SIGIR 2025 LiveRAG Challenge.

In this paper, we introduce DoTA-RAG (Dynamic-of-Thought Aggregation RAG), a retrieval-augmented generation system optimized for high-throughput, large-scale web knowledge indexes. Traditional RAG pipelines often suffer from high latency and limited accuracy over massive, diverse datasets. DoTA-RAG addresses these challenges with a three-stage pipeline: query rewriting, dynamic routing to specialized sub-indexes, and multi-stage retrieval and ranking. We further enhance retrieval by evaluating and selecting a superior embedding model, re-embedding the large FineWeb-10BT corpus. Moreover, we create a diverse Q&A dataset of 500 questions generated via the DataMorgana setup across a broad range of WebOrganizer topics and formats. DoTA-RAG improves the answer correctness score from 0.752 (baseline, using LiveRAG pre-built vector store) to 1.478 while maintaining low latency, and it achieves a 0.929 correctness score on the Live Challenge Day. These results highlight DoTA-RAG's potential for practical deployment in domains requiring fast, reliable access to large and evolving knowledge sources.

Historical maps are invaluable for analyzing long-term changes in transportation and spatial development, offering a rich source of data for evolutionary studies. However, digitizing and classifying road networks from these maps is often expensive and time-consuming, limiting their widespread use. Recent advancements in deep learning have made automatic road extraction from historical maps feasible, yet these methods typically require large amounts of labeled training data. To address this challenge, we introduce a novel framework that integrates deep learning with geoinformation, computer-based painting, and image processing methodologies. This framework enables the extraction and classification of roads from historical maps using only road geometries without needing road class labels for training. The process begins with training of a binary segmentation model to extract road geometries, followed by morphological operations, skeletonization, vectorization, and filtering algorithms. Synthetic training data is then generated by a painting function that artificially re-paints road segments using predefined symbology for road classes. Using this synthetic data, a deep ensemble is trained to generate pixel-wise probabilities for road classes to mitigate distribution shift. These predictions are then discretized along the extracted road geometries. Subsequently, further processing is employed to classify entire roads, enabling the identification of potential changes in road classes and resulting in a labeled road class dataset. Our method achieved completeness and correctness scores of over 94% and 92%, respectively, for road class 2, the most prevalent class in the two Siegfried Map sheets from Switzerland used for testing. This research offers a powerful tool for urban planning and transportation decision-making by efficiently extracting and classifying roads from historical maps.

Large language models (LLMs) can generate fluent reasoning traces that nevertheless lead to incorrect answers, making response-level uncertainty estimation important for abstention, human review, and adaptive compute allocation. Existing approaches generally fall into three categories: passive single-trace methods use token-level confidence signals, sampling-based methods compare multiple complete traces at higher generation cost, and active prefix-based methods probe partial traces to study answer stabilization or preference transitions. However, none actively re-elicits an answer from a completed reasoning trace to measure its consistency with and support for the original answer. To address this gap, we introduce Trace-Conditioned Answer Consistency (TrAC), a correctness-supervised uncertainty quantification framework that combines active and passive signals anchored to one completed reasoning trace. Its active component, Prefix-Conditioned Elicitation (PCE), re-elicits a short answer conditioned on the completed trace and represents both its consistency with the original answer and its token-level probabilistic support. Its passive component, Trace Uncertainty Profile (TUP), summarizes how token-level uncertainty evolves throughout the original generation without additional decoding. A lightweight head then integrates the two representations into a response-correctness score. Across five mathematical reasoning benchmarks and three LLM families, TrAC improves macro AUROC by 1.8% and reduces AURC by 3.4% relative to eight-sample self-consistency, while using one complete reasoning trace and a short cached answer probe. When eight samples are already available, augmenting sample consensus with re-elicitation further improves macro AUROC by 4.3% and reduces AURC by 8.3%, without additional full-trace generation.

When language models (LMs) are trained via reinforcement learning (RL) to generate natural language "reasoning chains", their performance improves on a variety of difficult question answering tasks. Today, almost all successful applications of RL for reasoning use binary reward functions that evaluate the correctness of LM outputs. Because such reward functions do not penalize guessing or low-confidence outputs, they often have the unintended side-effect of degrading calibration and increasing the rate at which LMs generate incorrect responses (or "hallucinate") in other problem domains. This paper describes RLCR (Reinforcement Learning with Calibration Rewards), an approach to training reasoning models that jointly improves accuracy and calibrated confidence estimation. During RLCR, LMs generate both predictions and numerical confidence estimates after reasoning. They are trained to optimize a reward function that augments a binary correctness score with a Brier score -- a scoring rule for confidence estimates that incentivizes calibrated prediction. We first prove that this reward function (or any analogous reward function that uses a bounded, proper scoring rule) yields models whose predictions are both accurate and well-calibrated. We next show that across diverse datasets, RLCR substantially improves calibration with no loss in accuracy, on both in-domain and out-of-domain evaluations -- outperforming both ordinary RL training and classifiers trained to assign post-hoc confidence scores. While ordinary RL hurts calibration, RLCR improves it. Finally, we demonstrate that verbalized confidence can be leveraged at test time to improve accuracy and calibration via confidence-weighted scaling methods. Our results show that explicitly optimizing for calibration can produce more generally reliable reasoning models.

Existing knowledge-grounded dialogue systems typically use finetuned versions of a pretrained language model (LM) and large-scale knowledge bases. These models typically fail to generalize on topics outside of the knowledge base, and require maintaining separate potentially large checkpoints each time finetuning is needed. In this paper, we aim to address these limitations by leveraging the inherent knowledge stored in the pretrained LM as well as its powerful generation ability. We propose a multi-stage prompting approach to generate knowledgeable responses from a single pretrained LM. We first prompt the LM to generate knowledge based on the dialogue context. Then, we further prompt it to generate responses based on the dialogue context and the previously generated knowledge. Results show that our knowledge generator outperforms the state-of-the-art retrieval-based model by 5.8% when combining knowledge relevance and correctness. In addition, our multi-stage prompting outperforms the finetuning-based dialogue model in terms of response knowledgeability and engagement by up to 10% and 5%, respectively. Furthermore, we scale our model up to 530 billion parameters and show that larger LMs improve the generation correctness score by up to 10%, and response relevance, knowledgeability and engagement by up to 10%. Our code is available at: https://github.com/NVIDIA/Megatron-LM.

Recent works on context and memory benchmarking have primarily focused on conversational instances but the need for evaluating memory in dynamic enterprise environments is crucial for its effective application. We introduce MEMTRACK, a benchmark designed to evaluate long-term memory and state tracking in multi-platform agent environments. MEMTRACK models realistic organizational workflows by integrating asynchronous events across multiple communication and productivity platforms such as Slack, Linear and Git. Each benchmark instance provides a chronologically platform-interleaved timeline, with noisy, conflicting, cross-referring information as well as potential codebase/file-system comprehension and exploration. Consequently, our benchmark tests memory capabilities such as acquistion, selection and conflict resolution. We curate the MEMTRACK dataset through both manual expert driven design and scalable agent based synthesis, generating ecologically valid scenarios grounded in real world software development processes. We introduce pertinent metrics for Correctness, Efficiency, and Redundancy that capture the effectiveness of memory mechanisms beyond simple QA performance. Experiments across SoTA LLMs and memory backends reveal challenges in utilizing memory across long horizons, handling cross-platform dependencies, and resolving contradictions. Notably, the best performing GPT-5 model only achieves a 60\% Correctness score on MEMTRACK. This work provides an extensible framework for advancing evaluation research for memory-augmented agents, beyond existing focus on conversational setups, and sets the stage for multi-agent, multi-platform memory benchmarking in complex organizational settings

Large Language Models (LLMs) and other automated techniques have been increasingly used to support software developers by generating software artifacts such as code snippets, patches, and comments. However, accurately assessing the correctness of these generated artifacts remains a significant challenge. On one hand, human evaluation provides high accuracy but is labor-intensive and lacks scalability. On the other hand, many automatic evaluation metrics are scalable and require minimal human effort, but they often fail to accurately reflect the actual correctness of generated software artifacts. In this paper, we present SE-Jury, the first evaluation metric for LLM-as-Ensemble-Judge specifically designed to accurately assess the correctness of generated software artifacts. SE-Jury first defines five distinct evaluation strategies, each implemented by an independent judge. A dynamic team selection mechanism then identifies the most appropriate subset of judges as a team to produce a final correctness score through ensembling. We evaluate SE-Jury across a diverse set of software engineering (SE) benchmarks that span three popular SE tasks: code generation, automated program repair, and code summarization. Results demonstrate that SE-Jury consistently achieves a higher correlation with human judgments, with improvements ranging from 29.6% to 140.8% over existing automatic metrics. SE-Jury reaches agreement levels with human annotators that are close to inter-annotator agreement in code generation and program repair. These findings underscore SE-Jury's potential as a scalable and reliable alternative to human evaluation in these SE tasks.

Retrieval-augmented generation (RAG) improves large language models (LLMs) by using external knowledge to guide response generation, reducing hallucinations. However, RAG, particularly multi-modal RAG, can introduce new hallucination sources: (i) the retrieval process may select irrelevant pieces (e.g., documents, images) as raw context from the database, and (ii) retrieved images are processed into text-based context via vision-language models (VLMs) or directly used by multi-modal language models (MLLMs) like GPT-4o, which may hallucinate. To address this, we propose a novel framework to evaluate the reliability of multi-modal RAG using two performance measures: (i) the relevancy score (RS), assessing the relevance of retrieved entries to the query, and (ii) the correctness score (CS), evaluating the accuracy of the generated response. We train RS and CS models using a ChatGPT-derived database and human evaluator samples. Results show that both models achieve ~88% accuracy on test data. Additionally, we construct a 5000-sample human-annotated database evaluating the relevancy of retrieved pieces and the correctness of response statements. Our RS model aligns with human preferences 20% more often than CLIP in retrieval, and our CS model matches human preferences ~91% of the time. Finally, we assess various RAG systems' selection and generation performances using RS and CS.

Large Language Models (LLMs) are fluent but prone to hallucinations, producing answers that appear plausible yet are unsupported by available evidence. This failure is especially problematic in high-stakes domains where decisions must be justified by verifiable information. We introduce EvidenceRL, a reinforcement learning framework that enforces evidence adherence during training. EvidenceRL scores candidate responses for grounding (entailment with retrieved evidence and context) and correctness (agreement with reference answers) and optimizes the generator using Group Relative Policy Optimization (GRPO). We evaluate across two high-stakes domains, cardiac diagnosis and legal reasoning, where EvidenceRL consistently improves evidence grounding and faithfulness without sacrificing task accuracy. On cardiac diagnosis, F1@3 increases from 37.0 to 54.5 on Llama-3.2-3B while grounding (G_{max}@3) rises from 47.6 to 78.2; hallucinations drop nearly 5times and evidence-supported diagnoses increase from 31.8\% to 61.6\%. On legal reasoning, EvidenceRL raises Faithfulness from 32.8\% to 67.6\% on Llama-3.1-8B, demonstrating consistent behavioral change across domains. Our code is open-sourced at https://github.com/Wizaaard/EvidenceRL.git.

Multi-hop reasoning over financial disclosures is often a retrieval problem before it becomes a reasoning or generation problem: relevant facts are dispersed across sections, filings, companies, and years, and LLMs often expend excessive tokens navigating noisy context. Without precise Knowledge Graph (KG)-guided selection of relevant context, even strong reasoning models either fail to answer or consume excessive tokens, whereas KG-linked evidence enables models to focus their reasoning on composing already retrieved facts. We present FinReflectKG - MultiHop, a benchmark built on FinReflectKG, a temporally indexed financial KG that links audited triples to source chunks from S&P 100 filings (2022-2024). Mining frequent 2-3 hop subgraph patterns across sectors (via GICS taxonomy), we generate financial analyst style questions with exact supporting evidence from the KG. A two-phase pipeline first creates QA pairs via pattern-specific prompts, followed by a multi-criteria quality control evaluation to ensure QA validity. We then evaluate three controlled retrieval scenarios: (S1) precise KG-linked paths; (S2) text-only page windows centered on relevant text spans; and (S3) relevant page windows with randomizations and distractors. Across both reasoning and non-reasoning models, KG-guided precise retrieval yields substantial gains on the FinReflectKG - MultiHop QA benchmark dataset, boosting correctness scores by approximately 24 percent while reducing token utilization by approximately 84.5 percent compared to the page window setting, which reflects the traditional vector retrieval paradigm. Spanning intra-document, inter-year, and cross-company scopes, our work underscores the pivotal role of knowledge graphs in efficiently connecting evidence for multi-hop financial QA. We also release a curated subset of the benchmark (555 QA Pairs) to catalyze further research.

Code generation is a core application of large language models (LLMs), yet LLMs still frequently fail on complex programming tasks. Given its success in mathematical reasoning, test-time scaling approaches such as Process Reward Model (PRM)-based Best-of-N selection offer a promising way to improve performance. However, existing PRMs remain ineffective for code generation due to the lack of meaningful step decomposition in code and the noise of Monte Carlo-estimated partial-solution correctness scores (rewards). To address these challenges, we propose FunPRM. FunPRM prompts LLMs to encourage modular code generation organized into functions, with functions treated as PRM reasoning steps. Furthermore, FunPRM introduces a novel meta-learning-based reward correction mechanism that leverages clean final-solution rewards obtained via a unit-test-based evaluation system to purify noisy partial-solution rewards. Experiments on LiveCodeBench and BigCodeBench demonstrate that FunPRM consistently outperforms existing test-time scaling methods across five base LLMs, notably achieving state-of-the-art performance on LiveCodeBench when combined with O4-mini. Furthermore, FunPRM produces code that is more readable and reusable for developers.

Reasoning abilities of LLMs have been a key focus in recent years. One challenging reasoning domain with interesting nuances is legal reasoning, which requires careful application of rules, and precedents while balancing deductive and analogical reasoning, and conflicts between rules. Although there have been a few works on using LLMs for legal reasoning, their focus has been on overall accuracy. In this paper, we dig deeper to do a step-by-step analysis and figure out where they commit errors. We use the college-level Multiple Choice Question-Answering (MCQA) task from the Civil Procedure dataset and propose a new error taxonomy derived from initial manual analysis of reasoning chains with respect to several LLMs, including two objective measures: soundness and correctness scores. We then develop an LLM-based automated evaluation framework to identify reasoning errors and evaluate the performance of LLMs. The computation of soundness and correctness on the dataset using the auto-evaluator framework reveals several interesting insights. Furthermore, we show that incorporating the error taxonomy as feedback in popular prompting techniques marginally increases LLM performance. Our work will also serve as an evaluation framework that can be used in detailed error analysis of reasoning chains for logic-intensive complex tasks.

RL for code correctness is now established: have the model generate a program, run it against hidden test cases, and reward solutions that pass. Extending this to code optimization seems straightforward: just add execution time to the reward. But in practice, once timing drives the reward, small problems in measurement noise, reward sparsity, or GRPO instability overwhelm the signal and make RL fail: generated solutions are barely faster, and more of them can fail. We make execution time learnable through three stages: (1) how code is tested, by building DMC-Optim with large optimization tests and a calibrated sandbox; (2) how speed is turned into reward, by composing correctness and speed in the RL environment and using an offline simulator to predict the most promising configurations; and (3) how the model learns from that reward, by adapting GRPO and evaluation to the sparser, noisier timed-execution setting. On DMC-Optim, the strongest optimization-aware configurations improve strict top-50% pass@1 from 18.0% to 31.3% on Qwen 2.5 7B and from 30.7% to 50.4% on CWM 32B. These gains further increase at stricter percentiles such as top-30%, with 125% relative improvement for CWM 32B, while preserving pure-correctness scores. When the timing sandbox is degraded, robust optimization RL reaches 100% to 200% improvement over standard RLVR, depending on the evaluation criterion. On LCB, CWM 32B wins up to 83% of median-sample speed comparisons against standard RLVR. Relative to the fastest correct human submissions per problem, it reaches about half the human rate of complexity-class improvements (14% vs. 28%).

Benchmarks for LLM-generated GPU kernels (KernelBench, TritonBench, GEAK) score correctness through fixed-shape, small-sample allclose-style checks. The number of inputs varies between benchmarks. The shape, dtype, and tolerance are fixed for each kernel. We test that oracle empirically. We construct a controlled corpus of 24 Triton and CPU stand-in kernels (15 correct controls and 9 LLM-style buggy variants seeded with documented transcription errors) and re-evaluate it under op-schema-aware seeded fuzzing with a high-precision (fp64) CPU reference and per-(op, dtype) absolute tolerances. The seeded oracle flags 9 of 9 buggy kernels and passes 15 of 15 correct controls, at zero precision cost on controls. We extend the corpus to 26 ops (adding a flash-attention pair) and re-run the same protocol on five GPU classes (RTX 3060, A10, L40S, A100 SXM4, H100 NVL). The verdicts are identical across all five GPUs: 10 of 10 illusions caught and 16 of 16 controls clean. The corpus result is about LLM-style transcription bugs that the allclose-on-one-shape oracle certifies as correct, not about the bug rate of any specific deployed LLM. Every flagged failure replays byte-for-byte from a stored seed.

Large language models excel on objectively verifiable tasks such as math and programming, where evaluation reduces to unit tests or a single correct answer. In contrast, real-world enterprise work is often subjective and context-dependent: success hinges on organizational goals, user intent, and the quality of intermediate artifacts produced across long, multi-tool workflows. We introduce LH-Bench, a three-pillar evaluation design that moves beyond binary correctness to score autonomous, long-horizon execution on subjective enterprise tasks. The pillars are: (i) expert-grounded rubrics that give LLM judges the domain context needed to score subjective work, (ii) curated ground-truth artifacts that enable stepwise reward signals (e.g., chapter-level annotation for content tasks), and (iii) pairwise human preference evaluation for convergent validation. We show that domain-authored rubrics provide substantially more reliable evaluation signals than LLM-authored rubrics (kappa = 0.60 vs. 0.46), and that human preference judgments confirm the same top-tier separation (p < 0.05), evidence that expert-grounded evaluation can scale without sacrificing reliability. We release public datasets and report results on two environments: Figma-to-code (33 real .fig tasks against the Figma API via MCP) and Programmatic content (41 courses comprising 183 individually-evaluated chapters on a course platform serving 30+ daily users).

Automating Text-to-Image (T2I) model evaluation is challenging; a judge model must be used to score correctness, and test prompts must be selected to be challenging for current T2I models but not the judge. We argue that satisfying these constraints can lead to benchmark drift over time, where the static benchmark judges fail to keep up with newer model capabilities. We show that benchmark drift is a significant problem for GenEval, one of the most popular T2I benchmarks. Although GenEval was well-aligned with human judgment at the time of its release, it has drifted far from human judgment over time -- resulting in an absolute error of as much as 17.7% for current models. This level of drift strongly suggests that GenEval has been saturated for some time, as we verify via a large-scale human study. To help fill this benchmarking gap, we introduce a new benchmark, GenEval 2, with improved coverage of primitive visual concepts and higher degrees of compositionality, which we show is more challenging for current models. We also introduce Soft-TIFA, an evaluation method for GenEval 2 that combines judgments for visual primitives, which we show is more well-aligned with human judgment and argue is less likely to drift from human-alignment over time (as compared to more holistic judges such as VQAScore). Although we hope GenEval 2 will provide a strong benchmark for many years, avoiding benchmark drift is far from guaranteed and our work, more generally, highlights the importance of continual audits and improvement for T2I and related automated model evaluation benchmarks.

Standard LLM benchmarks evaluate the assistant turn: the model generates a response to an input, a verifier scores correctness, and the analysis ends. This paradigm leaves unmeasured whether the LLM encodes any awareness of what follows the assistant response. We propose user-turn generation as a probe of this gap: given a conversation context of user query and assistant response, we let a model generate under the user role. If the model's weights encode interaction awareness, the generated user turn will be a grounded follow-up that reacts to the preceding context. Through experiments across 11 open-weight LLMs (Qwen3.5, gpt-oss, GLM) and 5 datasets (math reasoning, instruction following, conversation), we show that interaction awareness is decoupled from task accuracy. In particular, within the Qwen3.5 family, GSM8K accuracy scales from 41% (0.8B) to 96.8% (397B-A17B), yet genuine follow-up rates under deterministic generation remain near zero. In contrast, higher temperature sampling reveals interaction awareness is latent with follow up rates reaching 22%. Controlled perturbations validate that the proposed probe measures a real property of the model, and collaboration-oriented post-training on Qwen3.5-2B demonstrates an increase in follow-up rates. Our results show that user-turn generation captures a dimension of LLM behavior, interaction awareness, that is unexplored and invisible with current assistant-only benchmarks.

Large language models(LLMs) increasingly match expert performance on licensing examinations, yet routine clinical use remains limited because governance requires auditable reasoning, safety and ethics alignment, and resilience to adversarial misuse. Here we present SafeMed-R1, trained with a traceable Clinical Trust Signals(CTS) pipeline that links each reasoning instance to clinician rubric scores and edit histories, and aligned through safety and ethics supervision and red team stress testing. SafeMed-R1 attains a macro-averaged accuracy of 79.6% across clinical benchmarks. Under adversarial safety testing, it shows the lowest aggregated risk and reduces unsafe outputs by about 3 to 5% relative to its baseline. In a paired expert study of 30 medication safety vignettes, SafeMed-R1 matches PGY1 and PGY2 residents on medical correctness and scores higher for medication safety, guideline consistency, and clinical usefulness. Collectively, these results suggest that clinician-audited supervision provenance, together with domain-tailored safety and ethics alignment, can strengthen governance-relevant evidence without relying on inference-time retrieval or citation grounding.

Multimodal large language models (MLLMs) recently showed strong capacity in integrating data among multiple modalities, empowered by a generalizable attention architecture. Advanced methods predominantly focus on language-centric tuning while less exploring multimodal tokens mixed through attention, posing challenges in high-level tasks that require fine-grained cognition and emotion understanding. In this work, we identify the attention deficit disorder problem in multimodal learning, caused by inconsistent cross-modal attention and layer-by-layer decayed attention activation. To address this, we propose a novel attention mechanism, termed MOdular Duplex Attention (MODA), simultaneously conducting the inner-modal refinement and inter-modal interaction. MODA employs a correct-after-align strategy to effectively decouple modality alignment from cross-layer token mixing. In the alignment phase, tokens are mapped to duplex modality spaces based on the basis vectors, enabling the interaction between visual and language modality. Further, the correctness of attention scores is ensured through adaptive masked attention, which enhances the model's flexibility by allowing customizable masking patterns for different modalities. Extensive experiments on 21 benchmark datasets verify the effectiveness of MODA in perception, cognition, and emotion tasks. Source code and demo are available in https://zzcheng.top/MODA.

The HLTCOE LiveRAG submission utilized the GPT-researcher framework for researching the context of the question, filtering the returned results, and generating the final answer. The retrieval system was a ColBERT bi-encoder architecture, which represents a passage with many dense tokens. Retrieval used a local, compressed index of the FineWeb10-BT collection created with PLAID-X, using a model fine-tuned for multilingual retrieval. Query generation from context was done with Qwen2.5-7B-Instruct, while filtering was accomplished with m2-bert-80M-8k-retrieval. Up to nine passages were used as context to generate an answer using Falcon3-10B. This system placed 5th in the LiveRAG automatic evaluation for correctness with a score of 1.07.

The LiveRAG Challenge at SIGIR 2025, held between March and May 2025, provided a competitive platform for advancing Retrieval-Augmented Generation (RAG) technologies. Participants from academia and industry were invited to develop a RAG-based question-answering system using a fixed corpus (Fineweb-10BT) and a common open-source LLM (Falcon3-10B-Instruct). The goal was to facilitate challenging comparisons of retrieval and prompting strategies. During the Live Challenge Day, 70 teams from 27 different countries provided answers and supportive information to 500 unseen questions within a strict two-hour time window. Evaluation was conducted in two stages: first an automated LLM-as-a-judge approach was used to compute correctness and faithfulness score, then a manual review of top ranked submissions was conducted. The finalists were announced on June 12, 2025, with prizes awarded during the LiveRAG Workshop at SIGIR 2025 in Padua, Italy.

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.

Question answering (QA) can only make progress if we know if an answer is correct, but for many of the most challenging and interesting QA examples, current answer correctness (AC) metrics do not align with human judgments, particularly verbose, free form answers from large language models (LLM). There are two challenges: a lack of data and that models are too big. LLM based scorers correlate better with humans, but this expensive task has only been tested on limited QA datasets. We rectify these issues by providing clear guidelines for evaluating machine QA adopted from human QA contests. We also introduce Precise ANswer correctness Determination and Adjudication (PANDA), a small, efficient, deterministic AC classifier (812 KB) that more accurately evaluates answer correctness.

Reasoning prefixes shape the future trajectory of LLM problem solving, yet existing process reward models usually evaluate them through local step correctness. We argue that correctness is a useful but indirect proxy for the effect we ultimately care about: whether a prefix increases the probability of successful completion. We define this effect as prefix gain, the solve-rate improvement induced by conditioning lightweight student model group on a prefix, and use it to train a Prefix Utility Model (PUM) with a simple pairwise ranking objective. PUM learns outcome-grounded prefix utility and can score both complete trajectories and partial reasoning prefixes. Across Best-of-N selection, beam search, and reinforcement learning on mathematical reasoning, PUM provides a strong prefix-level supervision signal, especially when candidate pools are large, search budgets increase, or rule-based rewards are sparse. We release all data, models, and code at https://zhiqix.github.io/pum-project-page.

The ability to judge whether a caption correctly describes an image is a critical part of vision-language understanding. However, state-of-the-art models often misinterpret the correctness of fine-grained details, leading to errors in outputs such as hallucinating objects in generated captions or poor compositional reasoning. In this work, we explore Token-Level Confidence, or TLC, as a simple yet surprisingly effective method to assess caption correctness. Specifically, we fine-tune a vision-language model on image captioning, input an image and proposed caption to the model, and aggregate either algebraic or learned token confidences over words or sequences to estimate image-caption consistency. Compared to sequence-level scores from pretrained models, TLC with algebraic confidence measures achieves a relative improvement in accuracy by 10% on verb understanding in SVO-Probes and outperforms prior state-of-the-art in image and group scores for compositional reasoning in Winoground by a relative 37% and 9%, respectively. When training data are available, a learned confidence estimator provides further improved performance, reducing object hallucination rates in MS COCO Captions by a relative 30% over the original model and setting a new state-of-the-art.

Formal verification is the next frontier for ensuring the correctness of code generated by Large Language Models (LLMs). While methods that co-generate code and formal specifications in formal languages, like Dafny, can, in principle, prove alignment with user intent, progress is bottlenecked by specification quality evaluation. Current benchmarks rely on matching against ground-truth specifications, a manual and expertise-intensive process that has limited existing datasets to a few hundred simple problems and also suffers from a reliability issue. To address this, we introduce VeriEquivBench, a new benchmark with 2,389 complex algorithmic problems that probe the limitations of current models in both code generation and formal reasoning. Our evaluation framework replaces ground-truth matching with a formally grounded metric, the equivalence score, and rigorously verifies the quality of generated specifications and code. Our results show that generating formally verifiable code remains a profound challenge for state-of-the-art LLMs. This underscores both the difficulty of the task and the need for benchmarks like VeriEquivBench to drive progress toward scalable and reliable coding agents.

Visual captioning benchmarks have become outdated with the emergence of modern multimodal large language models (MLLMs), as the brief ground-truth sentences and traditional metrics fail to assess detailed captions effectively. While recent benchmarks attempt to address this by focusing on keyword extraction or object-centric evaluation, they remain limited to vague-view or object-view analyses and incomplete visual element coverage. In this paper, we introduce CAPability, a comprehensive multi-view benchmark for evaluating visual captioning across 12 dimensions spanning six critical views. We curate nearly 11K human-annotated images and videos with visual element annotations to evaluate the generated captions. CAPability stably assesses both the correctness and thoroughness of captions using F1-score. By converting annotations to QA pairs, we further introduce a heuristic metric, know but cannot tell (KT), indicating a significant performance gap between QA and caption capabilities. Our work provides the first holistic analysis of MLLMs' captioning abilities, as we identify their strengths and weaknesses across various dimensions, guiding future research to enhance specific aspects of capabilities.

Ensuring the correctness of code generated by Large Language Models (LLMs) presents a significant challenge in AI-driven software development. Existing approaches predominantly rely on black-box (closed-box) approaches that evaluate correctness post-generation, failing to utilize the rich insights embedded in the LLMs' internal states during code generation. In this paper, we introduce OPENIA, a novel white-box (open-box) framework that leverages these internal representations to assess the correctness of LLM-generated code. OPENIA systematically analyzes the intermediate states of representative open-source LLMs specialized for code, including DeepSeek-Coder, CodeLlama, and MagicCoder, across diverse code generation benchmarks. Our empirical analysis reveals that these internal representations encode latent information, which strongly correlates with the correctness of the generated code. Building on these insights, OPENIA uses a white-box/open-box approach to make informed predictions about code correctness, offering significant advantages in adaptability and robustness over traditional classification-based methods and zero-shot approaches. Experimental results demonstrate that OPENIA consistently outperforms baseline models, achieving higher accuracy, precision, recall, and F1-Scores with up to a 2X improvement in standalone code generation and a 46% enhancement in repository-specific scenarios. By unlocking the potential of in-process signals, OPENIA paves the way for more proactive and efficient quality assurance mechanisms in LLM-assisted code generation.

Large language models (LLMs) increasingly rely on explicit reasoning to solve coding tasks, yet evaluating the quality of this reasoning remains challenging. Existing reasoning evaluators are not designed for coding, and current benchmarks focus primarily on code generation, leaving other coding tasks largely unexplored. We introduce CodeRQ-Bench, the first benchmark for evaluating LLM reasoning quality across three coding task categories: generation, summarization, and classification. Using this benchmark, we analyze 1,069 mismatch cases from existing evaluators, identify five recurring limitations, and derive four design insights for reasoning evaluation in coding tasks. Guided by these insights, we propose VERA, a two-stage evaluator that combines evidence-grounded verification with ambiguity-aware score correction. Experiments on CodeRQ-Bench show that VERA consistently outperforms strong baselines across four datasets, improving AUCROC by up to 0.26 and AUPRC by up to 0.21. We release CodeRQ-Bench at https://github.com/MrLYG/CodeRQ-Bench, supporting future investigations.

Reliable assessment of the abilities of large audio language models (LALMs) is essential to advancing the state of the art. As benchmarks rapidly evolve to incorporate complex reasoning and subjective tasks, they increasingly necessitate open-ended responses from LALMs. We present Open-ended Response Correctness Assessment (ORCA) -- a reliable and lightweight model-based approach for answer correctness and disagreement modeling. We employ a three-stage annotation pipeline combining human judgment, structured feedback, and human-AI correction, yielding 9,663 annotations across 3,699 question-answer pairs from 15 LALMs on three audio understanding and reasoning benchmarks (achieving a Krippendorff's alpha of 0.82). Our experiments employing curriculum learning show that ORCA models achieve a Spearman correlation of 0.91 with average human correctness ratings on seen benchmarks and generalize to unseen benchmarks with a score of 0.85, outperforming several LLM judge baselines including Gemini 2.5 Flash. Furthermore, we demonstrate that ORCA's predicted variance correlates strongly with human disagreement, allowing it to effectively identify problematic benchmark items.

Evaluating the real-world applicability of large language models (LLMs) provides valuable insights for their development and use in software development tasks. Existing benchmarks often focus on standalone coding problems or specific libraries, overlooking multi-file, project-based scenarios and lacking a rigorous evaluation of consistency. The HackerRank-ASTRA Benchmark introduces project-based coding problems that mirror real-world scenarios. It evaluates model consistency through 32 runs (k = 32) and median standard deviation while incorporating taxonomy-level analysis to assess sub-skill capabilities. Initial evaluations on 65 problems show that the top three models -- o1, o1-preview, and Claude-3.5-Sonnet-1022 -- achieved comparable average scores of 75%, with no statistically significant differences in performance. Notably, Claude-3.5-Sonnet-1022 demonstrated the highest consistency across problems, with low variability (SD = 0.0497), which was statistically significant compared to other models, highlighting its reliability for real-world software development tasks.

Embedding-based code retrieval is a core component of coding agents and retrieval-augmented code generation, where retrieving correct code matters more than retrieving lexically similar code. Existing code-retrieval benchmarks do not plant controlled, execution-verified single-edit variants of each query's canonical implementation in the search pool, leaving the question of whether embeddings can functionally discriminate correct from near-clone-but-incorrect code unanswered in a retrieval setting. Resolving this requires a benchmark whose search pool itself contains the relevant counterfactuals -- execution-verified buggy variants near-identical to each canonical -- so that a retriever's rank ordering can be directly tested for functional discrimination rather than topical or identity overlap. We introduce ExecRetrieval, 939 Python tasks each paired with one execution-verified canonical implementation and up to four execution-verified buggy distractors, each generated by a mechanical mutation making a single targeted edit, and evaluate 23 dense embedding configurations plus BM25 under provider-native invocation with paired McNemar tests and query-level bootstrap intervals. With near-clone counterfactuals in the pool, the top hosted system reaches exec@10 = 1.00 but only exec@1 = 0.331; rank-1 misses are paired buggy variants 91.5-99.4% of the time across the four leading systems, and the canonical scores below at least one of its four paired distractors in 67-78% of queries on the leading systems. The full dataset, execution oracle, embedding matrices, environment snapshot, and pairwise statistical tests are released at the URL in Appendix D.

Model calibration aims to align confidence with prediction correctness. The Cross-Entropy (CE) loss is widely used for calibrator training, which enforces the model to increase confidence on the ground truth class. However, we find the CE loss has intrinsic limitations. For example, for a narrow misclassification, a calibrator trained by the CE loss often produces high confidence on the wrongly predicted class (e.g., a test sample is wrongly classified and its softmax score on the ground truth class is around 0.4), which is undesirable. In this paper, we propose a new post-hoc calibration objective derived from the aim of calibration. Intuitively, the proposed objective function asks that the calibrator decrease model confidence on wrongly predicted samples and increase confidence on correctly predicted samples. Because a sample itself has insufficient ability to indicate correctness, we use its transformed versions (e.g., rotated, greyscaled and color-jittered) during calibrator training. Trained on an in-distribution validation set and tested with isolated, individual test samples, our method achieves competitive calibration performance on both in-distribution and out-of-distribution test sets compared with the state of the art. Further, our analysis points out the difference between our method and commonly used objectives such as CE loss and mean square error loss, where the latters sometimes deviates from the calibration aim.

Large language models (LLMs) have demonstrated impressive capabilities in code generation, achieving high scores on benchmarks such as HumanEval and MBPP. However, these benchmarks primarily assess functional correctness and neglect broader dimensions of code quality, including security, reliability, readability, and maintainability. In this work, we systematically evaluate the ability of LLMs to generate high-quality code across multiple dimensions using the PythonSecurityEval benchmark. We introduce an iterative static analysis-driven prompting algorithm that leverages Bandit and Pylint to identify and resolve code quality issues. Our experiments with GPT-4o show substantial improvements: security issues reduced from >40% to 13%, readability violations from >80% to 11%, and reliability warnings from >50% to 11% within ten iterations. These results demonstrate that LLMs, when guided by static analysis feedback, can significantly enhance code quality beyond functional correctness.

Multi-agent reasoning systems often use agreement, confidence, or automated scores to decide which messages should shape a final answer. Such filtering assumes that a message likely to be correct is also worth keeping. Yet a wrong answer can contain a useful decomposition, constraint, or scientific principle. We test this distinction with Diverse Hypothesis Deliberation (DHD), a controlled measurement protocol that caches five independently generated messages and replays the same downstream solver, called the integrator, with each message available or hidden. The replay comparison measures a message's trajectory value: whether making the message available helps or harms subsequent reasoning. Across five mathematics and science benchmarks and two openly available model families, gpt-oss-120b and gemma-4-31B-it, wrong-helpful messages appear in every benchmark-model combination. Among wrong-answer messages that change final correctness, more than four in ten changes are helpful in each model. Controlled repeats show that the number of repeatable message effects is unlikely to arise from replay variation alone (p=0.0002). A focused intervention on repeatable wrong-helpful messages finds that the complete message works best, while retaining its reasoning preserves more success than retaining only its answer; the source of the complete-message advantage remains open. Within the same problem, repeated trajectory-value evidence also identifies a better keep-or-remove choice than answer correctness alone. Answer correctness is therefore informative but does not determine trajectory value. DHD measures this missing property and produces reusable labels for learning when agents should listen.

A binary decision task, like yes-no questions or answer verification, reflects a significant real-world scenario such as where users look for confirmation about the correctness of their decisions on specific issues. In this work, we observe that language models exhibit a negative bias in the binary decisions of complex reasoning tasks. Based on our observations and the rationale about attention-based model dynamics, we propose a negative attention score (NAS) to systematically and quantitatively formulate negative bias. Based on NAS, we identify attention heads that attend to negative tokens provided in the instructions as answer candidate of binary decisions, regardless of the question in the prompt, and validate their association with the negative bias. Additionally, we propose the negative attention score alignment (NASA) method, which is a parameter-efficient fine-tuning technique to address the extracted negatively biased attention heads. Experimental results from various domains of reasoning tasks and large model search space demonstrate that NASA significantly reduces the gap between precision and recall caused by negative bias while preserving their generalization abilities. Our codes are available at https://github.com/ysw1021/NASA.

Real-world data for classification is often labeled by multiple annotators. For analyzing such data, we introduce CROWDLAB, a straightforward approach to utilize any trained classifier to estimate: (1) A consensus label for each example that aggregates the available annotations; (2) A confidence score for how likely each consensus label is correct; (3) A rating for each annotator quantifying the overall correctness of their labels. Existing algorithms to estimate related quantities in crowdsourcing often rely on sophisticated generative models with iterative inference. CROWDLAB instead uses a straightforward weighted ensemble. Existing algorithms often rely solely on annotator statistics, ignoring the features of the examples from which the annotations derive. CROWDLAB utilizes any classifier model trained on these features, and can thus better generalize between examples with similar features. On real-world multi-annotator image data, our proposed method provides superior estimates for (1)-(3) than existing algorithms like Dawid-Skene/GLAD.

Unified multimodal models (UMMs) have shown impressive capabilities in generating natural images and supporting multimodal reasoning. However, their potential in supporting computer-use planning tasks, which are closely related to our lives, remain underexplored. Image generation and editing in computer-use tasks require capabilities like spatial reasoning and procedural understanding, and it is still unknown whether UMMs have these capabilities to finish these tasks or not. Therefore, we propose PlanViz, a new benchmark designed to evaluate image generation and editing for computer-use tasks. To achieve the goal of our evaluation, we focus on sub-tasks which frequently involve in daily life and require planning steps. Specifically, three new sub-tasks are designed: route planning, work diagramming, and web&UI displaying. We address challenges in data quality ensuring by curating human-annotated questions and reference images, and a quality control process. For challenges of comprehensive and exact evaluation, a task-adaptive score, PlanScore, is proposed. The score helps understanding the correctness, visual quality and efficiency of generated images. Through experiments, we highlight key limitations and opportunities for future research on this topic.

Discrete diffusion language models (dLLMs) expose a denoised solution at every step, which makes process reward model (PRM) guidance look like a way to spend compute at test time. We show that once denoising, PRM scoring, and outcome reward model (ORM) scoring are charged in the same budget of forward passes, its deterministic form loses to a much simpler baseline. Our PRMs score intermediate denoising states and are trained on the correctness of the final answer. On Dream-v0-Instruct-7B with 8 candidates per GSM8K problem, keeping the candidate with the highest PRM score at every scoring step reaches 65.18%, while independent sampling plus an ORM reranker trained for the task reaches 75.13%. The gap grows to 12.69 percentage points (pp) with 32 candidates, and is 9.85 pp on MATH and 12.16 pp on MBPP. We trace it to two separable failures. First, guidance prunes on a weak signal: on GSM8K, PRM ROC-AUC falls from 0.77 to 0.54 as the mask ratio rises, a decay that persists when states are relabeled with fresh rollouts, and pruning lowers the best accuracy reachable from the candidate pool from 81.05% for independent samples to 67.30%. Second, on GSM8K and MATH, the PRM is a poor final judge: a sequential Monte Carlo sampler at the same budget restores that ceiling to 77.89%, yet selecting with the PRM gives 65.48%, on par with deterministic guidance, while a PRM retrained on final states matches the ORM on identical candidates. MBPP separates the two: there the PRM reaches 65.47% when reranking finished programs, on par with the ORM, but 50.88% when it guides denoising. The results point to two targets for dLLM guidance: keep correct partial solutions alive through early denoising, and leave the final choice to a verifier trained on final states. We release the corpus of denoising states with outcome labels and evaluation toolkit for reproducible comparisons at matched compute.

LLM coding agents can generate working code, but their solutions often accumulate complexity, duplication, and architectural debt. Human developers address such issues through refactoring: behavior-preserving program transformations that improve structure and maintainability. We investigate whether agents (i) can execute refactorings reliably and (ii) identify the refactorings that human developers actually chose in real codebases. To this end, we construct CodeTaste, a benchmark mined from large multi-file open-source refactorings. To score solutions, we combine repository test suites that measure functional correctness with tailored static checks that verify removal of undesired and introduction of desired code patterns using dataflow reasoning. Our results show a clear gap: agents perform well at implementing refactorings that are specified in detail, but often fail to discover the human refactoring choices when given a focus area for changes. A propose-then-implement decomposition improves alignment, and selecting the best-aligned proposal before implementation can yield further gains. CodeTaste provides an evaluation target and a potential preference signal for aligning coding agents with human refactoring decisions in realistic codebases. We release the benchmark, leaderboard, and code.

Agent benchmarks often evaluate only final answers even when agents run on stateful runtimes. We argue this under-specifies what is being evaluated: the proper unit is a declared model-plus-runtime configuration whose failures can occur in evidence acquisition, runtime routing, safety boundaries, or repeated execution. We present ClawProBench, a trace-aware benchmark for runtime-native agent evaluation instantiated on OpenClaw, a live agent runtime with workspace tools and native surfaces for browsing, memory, messaging, scheduling, skills, and subagents. ClawProBench defines two tracks: a 102-scenario full profile with live workspace and native-runtime routing tasks, and a frozen 68-scenario holdout with closed-world JSON output contracts for robust ranking. Trials are scored from execution traces via a safety-gated formula combining correctness, process quality, and efficiency, preserving failure evidence for audit. Our anonymous artifact includes benchmark definitions, scoring code, manifests and sanitized traces. We evaluate 68 configurations on the full profile and 37 on holdout. The top safety-gated average trace score is 0.7671. Native-runtime tasks underperform workspace-live tasks (0.5238 vs. 0.6415). On holdout, pass@k-any outperforms strict three-trial pass (0.6638 vs. 0.2890), while full-profile and holdout rankings show weak alignment (Spearman 0.1300). Rankings based purely on correctness differ substantially from process-aware, safety-gated and strict-pass views. Final-answer leaderboards may hide native-surface weaknesses, one-off successes and trace-local agent failure modes.

What if a video generation model could not only imagine a plausible future, but the correct one, accurately reflecting how the world changes with each action? We address this question by presenting the Egocentric World Model (EgoWM), a simple, architecture-agnostic method that transforms any pretrained video diffusion model into an action-conditioned world model, enabling controllable future prediction. Rather than training from scratch, we repurpose the rich world priors of Internet-scale video models and inject motor commands through lightweight conditioning layers. This allows the model to follow actions faithfully while preserving realism and strong generalization. Our approach scales naturally across embodiments and action spaces, ranging from 3-DoF mobile robots to 25-DoF humanoids, where predicting egocentric joint-angle-driven dynamics is substantially more challenging. The model produces coherent rollouts for both navigation and manipulation tasks, requiring only modest fine-tuning. To evaluate physical correctness independently of visual appearance, we introduce the Structural Consistency Score (SCS), which measures whether stable scene elements evolve consistently with the provided actions. EgoWM improves SCS by up to 80 percent over prior state-of-the-art navigation world models, while achieving up to six times lower inference latency and robust generalization to unseen environments, including navigation inside paintings.

Small language models (SLMs) struggle with complex reasoning because exploration is expensive under tight compute budgets. We introduce Semantic Diversity-Exploration-Exploitation (SD-E^2), a reinforcement learning framework that makes exploration explicit by optimizing semantic diversity in generated reasoning trajectories. Using a frozen sentence-embedding model, SD-E^2 assigns a diversity reward that captures (i) the coverage of semantically distinct solution strategies and (ii) their average pairwise dissimilarity in embedding space, rather than surface-form novelty. This diversity reward is combined with outcome correctness and solution efficiency in a z-score-normalized multi-objective objective that stabilizes training. On GSM8K, SD-E^2 surpasses the base Qwen2.5-3B-Instruct and strong GRPO baselines (GRPO-CFL and GRPO-CFEE) by +27.4, +5.2, and +1.5 percentage points, respectively, while discovering on average 9.8 semantically distinct strategies per question. We further improve MedMCQA to 49.64% versus 38.37% for the base model and show gains on the harder AIME benchmark (1983-2025), reaching 13.28% versus 6.74% for the base. These results indicate that rewarding semantic novelty yields a more compute-efficient exploration-exploitation signal for training reasoning-capable SLMs. By introducing cognitive adaptation-adjusting the reasoning process structure rather than per-token computation-SD-E^2 offers a complementary path to efficiency gains in resource-constrained models.

Best-of-N (BoN) sampling, a common strategy for test-time scaling of Large Language Models (LLMs), relies on reward models to select the best candidate solution from multiple generations. However, traditional reward models often assign arbitrary and inconsistent scores, limiting their effectiveness. To address this, we propose a Pairwise Reward Model (Pairwise RM) combined with a knockout tournament for BoN sampling. Instead of assigning absolute scores, given one math problem, Pairwise RM evaluates two candidate solutions' correctness simultaneously. This approach eliminates the need for arbitrary scoring and enables cross-validation of solutions through parallel comparison. In the knockout tournament, Pairwise RM conducts pairwise comparisons between candidate solutions and eliminates the incorrect ones iteratively. We construct \ourdataset, a large-scale dataset of 443K pairwise comparisons derived from NumiaMath and annotated using gemini-1.5-flash, and train the Pairwise RM via supervised fine-tuning. Experiments on MATH-500 and the Olympiad Bench demonstrate significant improvements over traditional discriminative reward models. And a 40\% to 60\% relative improvement is achieved on the top 50\% challenging problems.

Large Language Models (LLMs) have catalyzed vibe coding, where users leverage LLMs to generate and iteratively refine code through natural language interactions until it passes their vibe check. Vibe check is tied to real-world human preference and goes beyond functionality: the solution should feel right, read cleanly, preserve intent, and remain correct. However, current code evaluation remains anchored to pass@k and captures only functional correctness, overlooking the non-functional instructions that users routinely apply. In this paper, we hypothesize that instruction following is the missing piece underlying vibe check that represents human preference in coding besides functional correctness. To quantify models' code instruction following capabilities with measurable signals, we present VeriCode, a taxonomy of 30 verifiable code instructions together with corresponding deterministic verifiers. We use the taxonomy to augment established evaluation suites, resulting in Vibe Checker, a testbed to assess both code instruction following and functional correctness. Upon evaluating 31 leading LLMs, we show that even the strongest models struggle to comply with multiple instructions and exhibit clear functional regression. Most importantly, a composite score of functional correctness and instruction following correlates the best with human preference, with the latter emerging as the primary differentiator on real-world programming tasks. Our work identifies core factors of the vibe check, providing a concrete path for benchmarking and developing models that better align with user preferences in coding.

Calibration of language models -- the alignment between expressed or implicit confidence and empirical correctness -- is a well-studied subfield within NLP. Methods to measure it already exist. The problem is adoption: outside this subfield, NLP research regularly introduces new models, datasets, and benchmarks without checking whether the model's confidence scores are meaningful. We argue that this adoption gap is a major obstacle to trustworthy LLM evaluation. Miscalibration causes problems in two distinct areas: at deployment, where overconfident mistakes cause real harm, and inside the research pipeline, where methods like LLM-as-a-judge, synthetic data generation, and active learning rely on calibrated confidence without verifying it. Standard calibration metrics only require two inputs per example: a confidence score and a correctness judgment. Most benchmarks in use today already provide both, meaning calibration can be reported immediately. For open-ended generation, however, defining these two inputs is still an open challenge. We argue that each NLP subfield should pair its main performance metric with a calibration score and call for treating calibration as an essential property of every model rather than a niche topic.

Integrating Large Language Models (LLMs) with Knowledge Graphs (KGs) results in complex systems with numerous hyperparameters that directly affect performance. While such systems are increasingly common in retrieval-augmented generation, the role of systematic hyperparameter optimization remains underexplored. In this paper, we study this problem in the context of Cognee, a modular framework for end-to-end KG construction and retrieval. Using three multi-hop QA benchmarks (HotPotQA, TwoWikiMultiHop, and MuSiQue) we optimize parameters related to chunking, graph construction, retrieval, and prompting. Each configuration is scored using established metrics (exact match, F1, and DeepEval's LLM-based correctness metric). Our results demonstrate that meaningful gains can be achieved through targeted tuning. While the gains are consistent, they are not uniform, with performance varying across datasets and metrics. This variability highlights both the value of tuning and the limitations of standard evaluation measures. While demonstrating the immediate potential of hyperparameter tuning, we argue that future progress will depend not only on architectural advances but also on clearer frameworks for optimization and evaluation in complex, modular systems.

Post-training for reasoning of large language models (LLMs) increasingly relies on verifiable rewards: deterministic checkers that provide 0-1 correctness signals. While reliable, such binary feedback is brittle--many tasks admit partially correct or alternative answers that verifiers under-credit, and the resulting all-or-nothing supervision limits learning. Reward models offer richer, continuous feedback, which can serve as a complementary supervisory signal to verifiers. We introduce HERO (Hybrid Ensemble Reward Optimization), a reinforcement learning framework that integrates verifier signals with reward-model scores in a structured way. HERO employs stratified normalization to bound reward-model scores within verifier-defined groups, preserving correctness while refining quality distinctions, and variance-aware weighting to emphasize challenging prompts where dense signals matter most. Across diverse mathematical reasoning benchmarks, HERO consistently outperforms RM-only and verifier-only baselines, with strong gains on both verifiable and hard-to-verify tasks. Our results show that hybrid reward design retains the stability of verifiers while leveraging the nuance of reward models to advance reasoning.

Large language models (LLMs) have demonstrated their remarkable capacity across a variety of tasks. However, reasoning remains a challenge for LLMs. To improve LLMs' reasoning ability, process supervision has proven to be better than outcome supervision. In this work, we study using Monte Carlo Tree Search (MCTS) to generate process supervision data with LLMs themselves for training them. We sample reasoning steps with an LLM and assign each step a score that captures its "relative correctness," and the LLM is then trained by minimizing weighted log-likelihood of generating the reasoning steps. This generate-then-train process is repeated iteratively until convergence.Our experimental results demonstrate that the proposed methods considerably improve the performance of LLMs on two mathematical reasoning datasets. Furthermore, models trained on one dataset also exhibit improved performance on the other, showing the transferability of the enhanced reasoning ability.

State-of-the-art text-to-speech (TTS) systems realize high naturalness in monolingual environments, synthesizing speech with correct multilingual accents (especially for Indic languages) and context-relevant emotions still poses difficulty owing to cultural nuance discrepancies in current frameworks. This paper introduces a new TTS architecture integrating accent along with preserving transliteration with multi-scale emotion modelling, in particularly tuned for Hindi and Indian English accent. Our approach extends the Parler-TTS model by integrating A language-specific phoneme alignment hybrid encoder-decoder architecture, and culture-sensitive emotion embedding layers trained on native speaker corpora, as well as incorporating a dynamic accent code switching with residual vector quantization. Quantitative tests demonstrate 23.7% improvement in accent accuracy (Word Error Rate reduction from 15.4% to 11.8%) and 85.3% emotion recognition accuracy from native listeners, surpassing METTS and VECL-TTS baselines. The novelty of the system is that it can mix code in real time - generating statements such as "Namaste, let's talk about <Hindi phrase>" with uninterrupted accent shifts while preserving emotional consistency. Subjective evaluation with 200 users reported a mean opinion score (MOS) of 4.2/5 for cultural correctness, much better than existing multilingual systems (p<0.01). This research makes cross-lingual synthesis more feasible by showcasing scalable accent-emotion disentanglement, with direct application in South Asian EdTech and accessibility software.

Recent studies have demonstrated the potential of Large Language Models (LLMs) in generating GPU Kernels. Current benchmarks focus on the translation of high-level languages into CUDA, overlooking the more general and challenging task of text-to-CUDA generation. Furthermore, given the hardware-specific and performance-critical features of GPU programming, accurately assessing the performance of LLM-generated GPU programs is nontrivial. In this work, we introduce CUDABench, a comprehensive benchmark designed to evaluate the text-to-CUDA capabilities of LLMs. First, we construct CUDABench-Set, which covers Breadth-Depth-Difficulty evaluation space in diverse application domains, including artificial intelligence, scientific computing, and data analytics, etc. Furthermore, we propose CUDABench-Score and Generative Verification Pipeline that assess (1) compilation correctness, (2) functional consistency through execution-based verification, and (3) a novel roofline-based metric, Performance-Score. Benchmarking state-of-the-art LLMs reveals insightful findings and challenges of text-to-CUDA, such as a notable mismatch between high compilation success rates and low functional correctness, a lack of domain-specific algorithmic knowledge, and suboptimal utilization of GPU hardware resources. Our benchmark is available at https://github.com/CUDA-Bench/CUDABench.

Self-evolution offers a scalable path to stronger reasoning: a pretrained language model improves itself with only minimal external supervision. Yet existing methods either depend on extensively curated or teacher-generated training data, or, when the generator runs unsupervised, reward it by a difficulty heuristic that need not improve the solver. We introduce INFUSER, an iterative co-training framework with two co-evolving roles: a Generator that drafts questions and reference golden answers from a pool of unstructured, automatically collected documents, and a Solver that improves by training on them. The solver is trained with standard correctness rewards against the generator-provided answers, while the generator is rewarded by an optimizer-aware influence score that measures whether each proposed question would actually improve the solver on the target distribution. Because this continuous, noisy influence score is poorly served by standard GRPO, we propose DuGRPO, a dual-normalized variant of GRPO, for generator training. Together, these turn the document pool into an adaptive curriculum that favors questions useful to the current solver, not just hard ones. On Qwen3-8B-Base, INFUSER outperforms strong self-evolution baselines with over 20% relative improvement on Olympiad and SuperGPQA benchmarks, and an 8B INFUSER co-evolving generator outperforms a frozen 32B thinking generator on math and coding. Ablations confirm each design choice is necessary, and two extensions, applying INFUSER to an instruction-finetuned anchor and augmenting it with rule-verifiable RLVR data, further demonstrate the flexibility and generalizability of the framework. Code is available at https://github.com/FFishy-git/INFUSER.

Production large language-model (LLM) agents are increasingly deployed not as lone problem-solvers but as managers: a main model creates specialized subagents, delegates work, and orchestrates their parallel, asynchronous returns through dynamic workflows. Whether one model can actually run such a team is largely unmeasured: existing benchmarks score a policy's own task-solving or a fixed multi-agent system's emergent behavior, but none isolate the management ability of the single LLM acting as leader. We introduce ClawArena-Team, a benchmark of 41 multi-turn, multimodal, multi-directory scenarios spanning 258 evaluation rounds and 72 staged updates that measures this management ability. The main agent is deliberately constrained: it natively perceives only text and directly accesses only part of the workspace. It commands a fixed, locally served subagent pool, so score differences reflect management skill, not raw capability. All scoring is execution-based with no LLM judge: an overall score -- the Subagent-Management Score (SMS) -- multiplies task correctness by a least-privilege and modality-routing factor. Across twelve proprietary, community-hosted, and self-hosted models, experiments show that the management bottleneck is privilege granting rather than perception (no model exceeds 50% workspace-permission precision); that cost and management quality are decoupled (API cost spans over 100 times while the overall score spans under 4 times, with the cheapest open models on the Pareto frontier); and that most leaderboard scores cluster within a 9.9-point band while orchestration behaviors diverge by more than an order of magnitude. Code is available at https://github.com/aiming-lab/ClawArena.

Since the rise of neural models of code that can generate long expressions and statements rather than a single next-token, one of the major problems has been reliably evaluating their generated output. In this paper, we propose CodeBERTScore: an automatic evaluation metric for code generation, which builds on BERTScore (Zhang et al., 2020). Instead of measuring exact token matching as BLEU, CodeBERTScore computes a soft similarity score between each token in the generated code and in the reference code, using the contextual encodings of large pretrained models. Further, instead of encoding only the generated tokens as in BERTScore, CodeBERTScore also encodes the programmatic context surrounding the generated code. We perform an extensive evaluation of CodeBERTScore across four programming languages. We find that CodeBERTScore achieves a higher correlation with human preference and with functional correctness than all existing metrics. That is, generated code that receives a higher score by CodeBERTScore is more likely to be preferred by humans, as well as to function correctly when executed. Finally, while CodeBERTScore can be used with a multilingual CodeBERT as its base model, we release five language-specific pretrained models to use with our publicly available code at https://github.com/neulab/code-bert-score . Our language-specific models have been downloaded more than 25,000 times from the Huggingface Hub.

This paper introduces VoxHakka, a text-to-speech (TTS) system designed for Taiwanese Hakka, a critically under-resourced language spoken in Taiwan. Leveraging the YourTTS framework, VoxHakka achieves high naturalness and accuracy and low real-time factor in speech synthesis while supporting six distinct Hakka dialects. This is achieved by training the model with dialect-specific data, allowing for the generation of speaker-aware Hakka speech. To address the scarcity of publicly available Hakka speech corpora, we employed a cost-effective approach utilizing a web scraping pipeline coupled with automatic speech recognition (ASR)-based data cleaning techniques. This process ensured the acquisition of a high-quality, multi-speaker, multi-dialect dataset suitable for TTS training. Subjective listening tests conducted using comparative mean opinion scores (CMOS) demonstrate that VoxHakka significantly outperforms existing publicly available Hakka TTS systems in terms of pronunciation accuracy, tone correctness, and overall naturalness. This work represents a significant advancement in Hakka language technology and provides a valuable resource for language preservation and revitalization efforts.

Large Reasoning Models have demonstrated remarkable performance with the advancement of test-time scaling techniques, which enhances prediction accuracy by generating multiple candidate responses and selecting the most reliable answer. While prior work has analyzed that internal model signals like confidence scores can partly indicate response correctness and exhibit a distributional correlation with accuracy, such distributional information has not been fully utilized to guide answer selection. Motivated by this, we propose DistriVoting, which incorporates distributional priors as another signal alongside confidence during voting. Specifically, our method (1) first decomposes the mixed confidence distribution into positive and negative components using Gaussian Mixture Models, (2) then applies a reject filter based on positive/negative samples from them to mitigate overlap between the two distributions. Besides, to further alleviate the overlap from the perspective of distribution itself, we propose SelfStepConf, which uses step-level confidence to dynamically adjust inference process, increasing the separation between the two distributions to improve the reliability of confidences in voting. Experiments across 16 models and 5 benchmarks demonstrate that our method significantly outperforms state-of-the-art approaches.

We introduce OmniInteract, a streaming benchmark for real-time omnimodal large language models evaluated through native online inference over audio-visual streams. Unlike offline video understanding or text-prompted streaming QA, OmniInteract preserves the original audio-visual stream and requires models to process it online, without access to future content. User queries and ambient sounds are embedded in the audio track, requiring models to detect multimodal triggers, decide when to respond, and answer while the stream unfolds. OmniInteract contains 250 videos with 1,430 temporally grounded response slots: 1,062 1Q1A slots across real-time, proactive, and nested scenarios, and 368 1QnA slots for continuous task monitoring and step guidance. Each slot includes a trigger, response window, and target answer. We evaluate response correctness, timing, invalid outputs, interruption handling, and context continuity using Interaction-Aware Quality-Timeliness F1, Interruption Diagnostic Suite, and Nested Chain Completion Score. Experiments show that current models remain weak in streaming interaction, with the best overall IA-QTF1 reaching only 0.368 and the best 1QnA IA-QTF1 only 0.052. Further study on mathematical reasoning in full-duplex settings shows that offline capability does not necessarily transfer to online interaction. Code and datasets will be made publicly accessible at https://github.com/Lucky-Lance/OmniInteract.

Collecting computer use data from human demonstrations is expensive and slow, motivating the need for scalable generation strategies. This requires two key ingredients: environments in which agents can act and verifiers that can judge whether their demonstrations succeeded. We introduce FaraGen1.5, a scalable data pipeline for computer use agents composed of three modular components: environments, solvers, and verifiers. FaraGen1.5 uses both live websites and synthetic environments that faithfully simulate domains gated by authentication or that require irreversible actions. It employs a solver harness that can be powered by multiple models, including strong frontier models such as GPT-5.4, and also incorporates a user simulator to enable multi-turn rollouts. Finally, FaraGen1.5 scores the resulting trajectories with three complementary verifiers covering task correctness, efficiency, and critical-point adherence. Using data produced by this pipeline, we train Fara1.5, a family of native computer use agents (CUAs) at three scales built on Qwen3.5 (4B, 9B, and 27B). To train these models, we employ a supervised finetuning (SFT) recipe that carefully balances data from FaraGen1.5 for broad coverage, specific high-value tasks, and target model deficiencies in an iterative approach. Each model sets a new state of the art for its size class on browser-use benchmarks: Fara1.5-9B reaches 63.4% on Online-Mind2Web and 86.6% on WebVoyager, while Fara1.5-27B achieves 72.3% on Online-Mind2Web, which is competitive with much larger proprietary systems.

Agentic coding harnesses - such as Agent-Skills, Superpowers, and Agent-Rigor - are increasingly deployed to augment underlying LLMs for real-world software engineering tasks. Existing benchmarks evaluate these agents almost exclusively on outcome correctness: whether generated code passes tests or resolves issues. We argue that this outcome-only lens is insufficient: an agent that arrives at a correct solution through reckless trial-and-error, without planning, verification, or graceful recovery, is fundamentally less reliable than one that follows sound engineering discipline. We introduce RigorBench, the first benchmark designed to measure process discipline in AI coding agents. RigorBench evaluates these harnesses across five pillars: Planning Fidelity, Verification Coverage, Recovery Efficiency, Abstention Quality, and Atomic Transition Integrity. A composite RigorScore aggregates these dimensions into a single metric via a weighted sum. We curate a suite of 30 tasks spanning five categories - Plan-Then-Build, Verify-Or-Die, Doom Loop Gauntlet, Know When to Fold, and Don't Break the Build-and evaluate leading harnesses in a controlled with/without experimental design against baseline coding assistants. Our results show that structured process discipline not only improves process quality scores by an average of 41% but also raises downstream outcome correctness by 17%, providing the first quantitative evidence that how agents code matters as much as what they produce. We release the full benchmark, scoring rubrics, and trajectory analysis tools as open-source artifacts.

Large language models (LLMs) have achieved expert-level performance on standardized examinations, yet multiple-choice accuracy poorly reflects real-world clinical utility and safety. As patients and clinicians increasingly use LLMs for guidance on complex conditions such as pancreatic cancer, evaluation must extend beyond general medical knowledge. Existing frameworks, such as HealthBench, rely on simulated queries and lack disease-specific depth. Moreover, high rubric-based scores do not ensure factual correctness, underscoring the need to assess hallucinations. We developed a human-in-the-loop pipeline to create expert rubrics for de-identified patient questions from the Pancreatic Cancer Action Network (PanCAN). The resulting benchmark, PanCanBench, includes 3,130 question-specific criteria across 282 authentic patient questions. We evaluated 22 proprietary and open-source LLMs using an LLM-as-a-judge framework, measuring clinical completeness, factual accuracy, and web-search integration. Models showed substantial variation in rubric-based completeness, with scores ranging from 46.5% to 82.3%. Factual errors were common, with hallucination rates (the percentages of responses containing at least one factual error) ranging from 6.0% for Gemini-2.5 Pro and GPT-4o to 53.8% for Llama-3.1-8B. Importantly, newer reasoning-optimized models did not consistently improve factuality: although o3 achieved the highest rubric score, it produced inaccuracies more frequently than other GPT-family models. Web-search integration did not inherently guarantee better responses. The average score changed from 66.8% to 63.9% for Gemini-2.5 Pro and from 73.8% to 72.8% for GPT-5 when web search was enabled. Synthetic AI-generated rubrics inflated absolute scores by 17.9 points on average while generally maintaining similar relative ranking.

Evaluating large language models (LLMs) for instrument control requires methods that go beyond standard, stateless algorithmic benchmarks, since the behavior of physical systems cannot be fully captured by unit tests alone. Here we introduce EnvTrace, a simulation-based method that evaluates execution traces to assess semantic code equivalence. EnvTrace is demonstrated with a beamline control-logic digital twin to facilitate the evaluation of instrument control code, with the digital twin itself also enabling the pre-execution validation of live experiments. Over 30 LLMs were evaluated using trace alignment to generate a multi-faceted score for functional correctness across key behavioral dimensions, showing that many top-tier models can approach human-level performance in rapid control-code generation. This is a first step toward a broader vision where LLMs and digital twins work symbiotically: LLMs providing intuitive control and agentic orchestration, and digital twins offering safe and high-fidelity environments, paving the way towards autonomous embodied AI.

Frontier coding agents solve complex tasks when given complete context but collapse when specifications are incomplete or ambiguous. The bottleneck is not raw capability, but judgment: knowing when to act autonomously and when to ask for help. Current benchmarks are blind to this failure mode. They supply unambiguous detailed instructions and solely reward execution correctness, so an agent that makes a lucky guess for a missing requirement will score identically to one that would have asked to be certain. We present HiL-Bench (Human-in-the-Loop Benchmark) to measure this selective escalation skill. Each task contains human-validated blockers (missing information, ambiguous requests, contradictory information) that surface only through progressive exploration, not upfront inspection. Our core metric, Ask-F1, the harmonic mean of question precision and blocker recall, captures the tension between over-asking and silent guessing; its structure architecturally prevents gaming through question spam. Evaluation across SWE and text-to-SQL domains reveals a large universal judgment gap: no frontier model recovers more than a fraction of its full-information performance when deciding whether to ask. Failure analysis identifies three key help-seeking patterns: overconfident wrong beliefs with no gap detection; high uncertainty detection yet persistent errors; broad, imprecise escalation without self-correction. These consistent patterns confirm poor help-seeking is a model-level flaw, not task-specific. RL training on shaped Ask-F1 reward shows judgment is trainable: a 32B model improves both help-seeking quality and task pass rate, with gains that transfer across domains. The model does not learn domain-specific heuristics for when to ask; it learns to detect unresolvable uncertainty and act on it.

Sparse Autoencoders (SAEs) decompose large language model representations into interpretable features, but how these features interact under uncertainty remains poorly understood. We introduce Feature Rivalry -- negatively correlated SAE feature pairs -- and study whether rivalry serves as a mechanistic signature of model uncertainty in Gemma-2-2B using Gemma Scope SAEs. Through a controlled within-domain experiment on PopQA split by response entropy, we find that high-entropy questions produce significantly stronger feature rivalry at layers 0 and 12 relative to low-entropy questions (p=5.3x10^-26 and p=5.8x10^-5 respectively), localizing uncertainty to specific processing stages in the residual stream. We then test whether rivalry is causally upstream of model outputs via activation steering along rivalry axes -- finding that steering along the rivalry direction (vec_A - vec_B) causes more output changes than random directions at low steering multipliers across 15 of 20 rival feature pairs. Finally, a per-prompt rivalry score derived from pairwise cosine similarities of active SAE feature decoder vectors predicts answer correctness (AUROC=0.689), approaching but not matching softmax confidence (AUROC=0.808).

LLM-based agents for GPU kernel generation are advancing rapidly, yet their progress is fundamentally constrained by the benchmarks they optimize against. Existing benchmarks are poorly aligned with production inference frameworks: they evaluate kernels on a single GPU with synthetic inputs, ignore the surrounding compilation stack, and reward replicating known optimizations rather than discovering new ones. The resulting reward signals are misleading: agents learn to generate kernels that score well in sandboxes but introduce interface incompatibilities, compilation-stack conflicts, and silent correctness degradation when integrated into real systems. We introduce FastKernels, a kernel benchmark built around a minimal set of 46 representative architectures spanning 8 categories, whose kernels collectively subsume those of 96.2% (409/425) of HuggingFace Transformers architectures. FastKernels doubles as a minimalistic, production-grade inference framework that runs at parity with hardened systems such as vLLM and SGLang on mainstream LLM serving and substantially exceeds upstream references on under-served architectures; each task's interface mirrors the corresponding module in the state-of-the-art library for its architecture family, enabling direct deployment of optimized kernels into production codebases. Evaluating state-of-the-art kernel agents on FastKernels, we find that even the strongest agent achieves only 0.94times aggregate speedup over production baselines, with weaker agents at 0.78times and 0.53times -- confirming that benchmark-production misalignment is a critical bottleneck for the field. We release FastKernels as a stepping stone toward kernel agents whose benchmark gains translate directly into production throughput improvements. Code is available at https://github.com/Snowflake-AI-Research/fastkernels

Layout generation is a task to synthesize a harmonious layout with elements characterized by attributes such as category, position, and size. Human designers experiment with the placement and modification of elements to create aesthetic layouts, however, we observed that current discrete diffusion models (DDMs) struggle to correct inharmonious layouts after they have been generated. In this paper, we first provide novel insights into layout sticking phenomenon in DDMs and then propose a simple yet effective layout-assessment module Layout-Corrector, which works in conjunction with existing DDMs to address the layout sticking problem. We present a learning-based module capable of identifying inharmonious elements within layouts, considering overall layout harmony characterized by complex composition. During the generation process, Layout-Corrector evaluates the correctness of each token in the generated layout, reinitializing those with low scores to the ungenerated state. The DDM then uses the high-scored tokens as clues to regenerate the harmonized tokens. Layout-Corrector, tested on common benchmarks, consistently boosts layout-generation performance when in conjunction with various state-of-the-art DDMs. Furthermore, our extensive analysis demonstrates that the Layout-Corrector (1) successfully identifies erroneous tokens, (2) facilitates control over the fidelity-diversity trade-off, and (3) significantly mitigates the performance drop associated with fast sampling.

Language models (LMs) should provide reliable confidence estimates to help users detect mistakes in their outputs and defer to human experts when necessary. Asking a language model to assess its confidence ("Score your confidence from 0-1.") is a natural way of evaluating its uncertainty. However, models struggle to provide absolute assessments of confidence (i.e. judging confidence in answering a question independent of other questions) and the coarse-grained scores they produce are not useful for evaluating the correctness of their answers. We propose relative confidence estimation, where we match up questions against each other and ask the model to make relative judgments of confidence ("Which question are you more confident in answering correctly?"). Treating each question as a "player" in a series of matchups against other questions and the model's preferences as match outcomes, we can use rank aggregation methods like Elo rating and Bradley-Terry to translate the model's confidence preferences into confidence scores. We evaluate relative confidence estimation against absolute confidence estimation and self-consistency confidence methods on five state-of-the-art LMs -- GPT-4, GPT-4o, Gemini 1.5 Pro, Claude 3.5 Sonnet, and Llama 3.1 405B -- across 14 challenging STEM, social science, and commonsense reasoning question answering tasks. Our results demonstrate that relative confidence estimation consistently provides more reliable confidence scores than absolute confidence estimation, with average gains of 3.5% in selective classification AUC over direct absolute confidence estimation methods and 1.7% over self-consistency approaches across all models and datasets.

This paper presents an eight-week observational comparison of 68 single-file HTML generations collected across 17 public experiments in the "HTML AI Battle" project between December 10, 2025 and February 4, 2026. Four reasoning model families, GPT, Gemini, Grok, and Claude, were compared under a fixed public-interface protocol with no custom instructions, no personality tuning, and no repair prompts. Each output was evaluated from a rendered browser video using human scores and a Gemini LLM-as-a-judge layer for prompt adherence, functional correctness, and UI quality, then packaged into a standardized social-media protocol spanning X (Twitter), TikTok, and YouTube. The tracker was also used for two supervised predictive analyses: an experiment-level model for 24-hour X impressions and a generation-level model for HTML verbosity. Under this protocol, Claude was the strongest and most consistent family, leading mean performance and winning 9/17 prompts under the primary human weighted score. Longer measured reasoning time was not associated with higher quality overall. Gemini as a judge was significantly more lenient than the human evaluator on functional correctness and overall performance, while stable self-favoring bias remained unresolved. The exploratory X-impressions model remained weak under post-screen cross-validation (MAE = 46,874, R^2 = -0.377), whereas the HTML-lines model performed better, with a model-family-only baseline outperforming prompt-aware alternatives (MAE = 135.2, R^2 = 0.576). Overall, selected pre-publication technical/audio variables were not sufficient to predict 24-hour X reach, while code verbosity was driven much more by model family than by prompt wording. The comparisons remain observational and are limited by public-interface drift, access-path differences, and one primary human scorer.

Video reasoning language models implicitly assume that every input frame is equally reliable. This leads to what we term the Blind Trust Problem: under realistic perturbations such as motion blur, glare, or occlusion, frontier video reasoning models can suffer 15-30%p accuracy drops on real-world embodied benchmarks, while remaining unaware that their visual evidence has been degraded. To address this challenge, we propose Robust-TO, an agentic video understanding framework that explicitly integrates per-frame trustworthiness into every stage of reasoning. Robust-TO organizes heterogeneous visual perception tools under a unified evidence interface. Each tool receives a sub-query derived from the original question and a set of trustworthy frames selected by the reliability-relevance score. It returns evidence in a shared format: a concrete prediction (e.g., a bounding box, motion trajectory, recognized text, or action label), temporal grounding, and a calibrated reliability score. During reasoning, these calibrated scores guide evidence weighting in a three-tier synthesis process (high/medium/low) and define a confidence-cost GRPO reward that jointly optimizes correctness, evidence reliability, and efficiency. On two video reasoning benchmarks spanning eight tasks, Robust-TO achieves 56.4% average accuracy on clean inputs, surpassing the strongest open-source baseline by 10.6%p and outperforming Gemini-2.5-Pro (46.2%). Under five realistic corruption types, Robust-TO maintains 54.3% average accuracy, 5.8%p above the strongest open-source baseline, while exhibiting the smallest clean-to-corrupted accuracy drop among all compared methods.

We propose two new methods for multi-document financial question answering. First, a method that uses semantic tagging, and then, queries the index to get the context (RAG_SEM). And second, a Knowledge Graph (KG_RAG) based method that uses semantic tagging, and, retrieves knowledge graph triples from a graph database, as context. KG_RAG uses knowledge graphs constructed using a small model that is fine-tuned using knowledge distillation using a large teacher model. The data consists of 18 10K reports of Apple, Microsoft, Alphabet, NVIDIA, Amazon and Tesla for the years 2021, 2022 and 2023. The list of questions in the data consists of 111 complex questions including many esoteric questions that are difficult to answer and the answers are not completely obvious. As evaluation metrics, we use overall scores as well as segmented scores for measurement including the faithfulness, relevance, correctness, similarity, an LLM based overall score and the rouge scores as well as a similarity of embeddings. We find that both methods outperform plain RAG significantly. KG_RAG outperforms RAG_SEM in four out of nine metrics.

The webpage-to-code task requires models to understand visual representations of webpages and generate corresponding code. However, existing benchmarks primarily focus on static screenshot-to-code tasks, thereby overlooking the dynamic interactions fundamental to real-world web applications. To address this limitation, this paper introduces IWR-Bench, a novel benchmark for evaluating the capabilities of Large Vision-Language Models (LVLMs) in interactive webpage reconstruction from video. IWR-Bench comprises 113 meticulously curated tasks from 100 real-world websites, with 1,001 actions and featuring diverse interaction complexities (e.g., web games), visual styles, and domains. Aligning with standard web development practices, each task includes not only user interaction videos but also all crawled static assets (e.g., images, videos). This benchmark evaluates models on two fundamental challenges: comprehensive multi-modal reasoning to infer interaction logic from video and assets, and advanced code generation to translate this logic into functional code. An agent-as-a-judge framework with a comprehensive metric system automatically assesses the functional correctness and visual fidelity of generated webpages. Extensive experiments on 28 LVLMs reveal a significant challenge: the best model achieves an overall score of only 36.35%, as functional correctness (24.39% IFS) lags significantly behind visual fidelity (64.25% VFS). These results highlight critical limitations in current models' ability to reason about temporal dynamics and synthesize event-driven logic, establishing IWR-Bench as a challenging frontier for vision-language research. The benchmark and evaluation code will be made publicly available. Code is available at https://github.com/L-O-I/IWR-Bench.

Parallel test-time scaling typically trains separate generation and verification models, incurring high training and inference costs. We propose Advantage Decoupled Preference Optimization (ADPO), a unified reinforcement learning framework that jointly learns answer generation and self-verification within a single policy. ADPO introduces two innovations: a preference verification reward improving verification capability and a decoupled optimization mechanism enabling synergistic optimization of generation and verification. Specifically, the preference verification reward computes mean verification scores from positive and negative samples as decision thresholds, providing positive feedback when prediction correctness aligns with answer correctness. Meanwhile, the advantage decoupled optimization computes separate advantages for generation and verification, applies token masks to isolate gradients, and combines masked GRPO objectives, preserving generation quality while calibrating verification scores. ADPO achieves up to +34.1% higher verification AUC and -53.5% lower inference time, with significant gains of +2.8%/+1.4% accuracy on MathVista/MMMU, +1.9 cIoU on ReasonSeg, and +1.7%/+1.0% step success rate on AndroidControl/GUI Odyssey.

Code generation models have shown significant potential for programming tasks. However, existing training methods like supervised fine-tuning face key limitations: they do not effectively teach models to prioritize correct over incorrect solutions in ambiguous situations, nor do they effectively optimize the runtime efficiency of the generated code. To address these challenges, we propose CodeDPO, a framework that integrates preference learning into code generation to improve two key code preference factors: code correctness and efficiency. CodeDPO employs a novel dataset construction method, utilizing a self-generation-and-validation mechanism that simultaneously generates and evaluates code and test cases. The underlying assumption is that test cases executable by multiple code snippets provide more reliable validation, and code that passes more tests is more likely to be correct. Through this self-validation process, our PageRank-inspired algorithm iteratively updates the ranking score of each code snippet, ultimately creating a code preference optimization dataset based on correctness and efficiency. CodeDPO is flexible and scalable, generating diverse preference optimization data without depending on external resources. Through comprehensive evaluations of five widely used benchmarks, CodeDPO demonstrates significant improvements in correctness and efficiency compared to existing methods. Our experiments prove that CodeDPO enhances the capabilities of LLMs in code generation and provides a robust foundation for conducting code preference optimization in more complex and challenging real-world scenarios.

In electronic design, engineers often manually search through extensive documents to retrieve component parameters required for constructing SPICE models, a process that is both labor-intensive and time-consuming. To address this challenge, we present an automated framework called D2S-FLOW that leverages large language models (LLMs) to extract electrical parameters from datasheets and generate SPICE models with high precision and efficiency, significantly reducing the need for manual intervention. Unlike traditional RAG systems, D2S-FLOW employs a workflow to enhance precision in handling unstructured documents and inconsistent naming conventions through three innovative mechanisms: Attention-Guided Document Focusing (AGDF), Hierarchical Document-Enhanced Retrieval (HDER), and Heterogeneous Named Entity Normalization (HNEN). AGDF narrows retrieval to user-selected documents, HDER utilizes document structure for precise parameter localization, and HNEN standardizes terminology via semantic inference. Experimental results demonstrate that the framework achieves an Exact Match (EM) of 0.86, an F1 score of 0.92, and an Exact Correctness (EC) of 0.96, outperforming the strongest baseline by 19.4%, 5.7%, and 13.1%, respectively. Additionally, it reduces API token consumption by 38% and minimizes the irrelevant information ratio to 4%, showcasing substantial improvements in resource efficiency. This research provides an effective automated solution for circuit design.

We introduce GraphNet, a dataset of 2.7K real-world deep learning computational graphs with rich metadata, spanning six major task categories across multiple deep learning frameworks. To evaluate tensor compiler performance on these samples, we propose the benchmark metric Speedup Score S(t), which jointly considers runtime speedup and execution correctness under tunable tolerance levels, offering a reliable measure of general optimization capability. Furthermore, we extend S(t) to the Error-aware Speedup Score ES(t), which incorporates error information and helps compiler developers identify key performance bottlenecks. In this report, we benchmark the default tensor compilers, CINN for PaddlePaddle and TorchInductor for PyTorch, on computer vision (CV) and natural language processing (NLP) samples to demonstrate the practicality of GraphNet. The full construction pipeline with graph extraction and compiler evaluation tools is available at https://github.com/PaddlePaddle/GraphNet .

Evaluating whether multimodal large language models truly understand long-form scientific papers remains challenging: answer-only metrics and synthetic "Needle-In-A-Haystack" tests often reward answer matching without requiring a causal, evidence-linked reasoning trace in the document. We propose the "Fish-in-the-Ocean" (FITO) paradigm, which requires models to construct explicit cross-modal evidence chains within native scientific documents. To operationalize FITO, we build SIN-Data, a scientific interleaved corpus that preserves the native interleaving of text and figures. On top of it, we construct SIN-Bench with four progressive tasks covering evidence discovery (SIN-Find), hypothesis verification (SIN-Verify), grounded QA (SIN-QA), and evidence-anchored synthesis (SIN-Summary). We further introduce "No Evidence, No Score", scoring predictions when grounded to verifiable anchors and diagnosing evidence quality via matching, relevance, and logic. Experiments on eight MLLMs show that grounding is the primary bottleneck: Gemini-3-pro achieves the best average overall score (0.573), while GPT-5 attains the highest SIN-QA answer accuracy (0.767) but underperforms on evidence-aligned overall scores, exposing a gap between correctness and traceable support.

Visual thinking should not only sound right; it should show its evidence. While recent vision-language models (VLMs) can produce natural-language reasoning traces, these traces often leave the supporting image regions implicit, making them hard to verify and difficult to supervise. We introduce visually grounded thinking, a reasoning process in which models interleave natural-language thoughts with explicit point or box groundings of the visual evidence used at each step. This lets the model express intermediate reasoning in language while grounding key objects in the image regions they refer to. To train this behavior, we construct a scalable synthesis pipeline that distills correct visual reasoning traces, extracts the visual objects required by the traces, grounds them with a SAM3-based agent, and derives aligned point and box supervision from the resulting masks. We further propose grounding-aware reinforcement learning, which combines answer correctness rewards with dense grounding rewards that score whether generated object references match the correct image evidence. Across two counting benchmarks and four spatial reasoning benchmarks, adding visually grounded thinking to Gemma3-4B-IT consistently improves performance over the original model and the non-grounded thinking baseline. On spatial reasoning, the visually grounded thinking 4B models match, and in some cases surpass, Gemma3-27B-IT from the same model family. Our analysis shows that point grounding is well suited to counting, while box grounding benefits most from explicit grounding rewards on spatial tasks. Overall, our results show that VLMs think better when their intermediate thoughts are tied to the image regions that make them true.

Automating AI research holds immense potential for accelerating scientific progress, yet current AI agents struggle with the complexities of rigorous, end-to-end experimentation. We introduce EXP-Bench, a novel benchmark designed to systematically evaluate AI agents on complete research experiments sourced from influential AI publications. Given a research question and incomplete starter code, EXP-Bench challenges AI agents to formulate hypotheses, design and implement experimental procedures, execute them, and analyze results. To enable the creation of such intricate and authentic tasks with high-fidelity, we design a semi-autonomous pipeline to extract and structure crucial experimental details from these research papers and their associated open-source code. With the pipeline, EXP-Bench curated 461 AI research tasks from 51 top-tier AI research papers. Evaluations of leading LLM-based agents, such as OpenHands and IterativeAgent on EXP-Bench demonstrate partial capabilities: while scores on individual experimental aspects such as design or implementation correctness occasionally reach 20-35%, the success rate for complete, executable experiments was a mere 0.5%. By identifying these bottlenecks and providing realistic step-by-step experiment procedures, EXP-Bench serves as a vital tool for future AI agents to improve their ability to conduct AI research experiments. EXP-Bench is open-sourced at https://github.com/Just-Curieous/Curie/tree/main/benchmark/exp_bench.

Text-to-SQL, the task of translating natural language questions into SQL queries, has significantly advanced with the introduction of Large Language Models (LLMs), broadening database accessibility for a wide range of users. Despite substantial progress in generating valid SQL, current LLMs still struggle with complex queries that require precise alignment between user intent and the database schema. To mitigate this, test-time strategies such as Best-of-N (BoN) and Majority Voting (Maj) are often employed, based on the assumption that LLMs can generate correct answers but may require multiple attempts. However, these methods rely on surface-level heuristics, selecting either the syntactically correct query through execution-based BoN (ex-BoN) or the most frequently generated query with Maj. Recently, Outcome Reward Models (ORMs), which assign utility scores to generated outputs based on semantic correctness, have emerged as a promising approach for better aligning model predictions with user intent. Nevertheless, their application to Text-to-SQL remains largely underexplored. In this work, we evaluate ORMs as an effective heuristic for BoN, compare them with ex-BoN and Maj, and introduce a framework for training ORMs for the Text-to-SQL task. We evaluate our ORMs on the BIRD and SPIDER benchmarks, finetuning various open-source LLMs, including the Qwen2, Granite3, and Llama3 model families. Our results show that ORMs outperform ex-BoN and Maj, achieving execution accuracy gains of +4.33% (BIRD) and +2.10% (Spider) over ex-BoN, and +2.91% (BIRD) and +0.93% (Spider) over Maj. We further demonstrate that finetuning models already aligned with SQL generation, such as OmniSQL, yields superior ORM performance. Additionally, we observe that ORMs achieve competitive results on simple queries and benefit more from an increased number of candidates compared to ex-BoN and Maj.

Recent video multimodal large language models achieve impressive results across various benchmarks. However, current evaluations suffer from two critical limitations: (1) inflated scores can mask deficiencies in fine-grained visual understanding and reasoning, and (2) answer correctness is often measured without verifying whether models identify the precise spatio-temporal evidence supporting their predictions. To address this, we present VideoZeroBench, a hierarchical benchmark designed for challenging long-video question answering that rigorously verifies spatio-temporal evidence. It comprises 500 manually annotated questions across 13 domains, paired with temporal intervals and spatial bounding boxes as evidence. To disentangle answering generation, temporal grounding, and spatial grounding, we introduce a five-level evaluation protocol that progressively tightens evidence requirements. Experiments show that even Gemini-3-Pro correctly answers fewer than 17% of questions under the standard end-to-end QA setting (Level-3). When grounding constraints are imposed, performance drops sharply: No model exceeds 1% accuracy when both correct answering and accurate spatio-temporal localization are required (Level-5), with most failing to achieve any correct grounded predictions. These results expose a significant gap between surface-level answer correctness and genuine evidence-based reasoning, revealing that grounded video understanding remains a bottleneck for long-video QA. We further analyze performance across minimal evidence spans, atomic abilities, and inference paradigms, providing insights for future research in grounded video reasoning. The benchmark and code will be made publicly available.

We introduce Sci-VBench, a comprehensive benchmark for evaluating knowledge- and reasoning-intensive video generation across scientific domains. It contains 1,253 expert-annotated examples spanning 60 subjects across four core disciplines: Natural Science, Healthcare, Humanities & Social Sciences, and Engineering. Each example requires models to generate temporally rich videos that demand scientific reasoning and knowledge-grounded synthesis, going beyond surface-level visual plausibility. We further establish a rubric-based evaluation protocol. Our analysis shows that, under this protocol, both non-expert human evaluators and MLLM-as-Judge systems can achieve relatively high agreement with expert judgments, supporting reproducible evaluation at scale. We benchmark 16 frontier proprietary and open-source models and find that, while automatic perceptual-quality scores cluster tightly across systems, performance on Prompt Grounding and Scientific and Causal Correctness varies substantially, with a pronounced proprietary-open-source gap. These findings show that advances in visual realism have not yet translated into reliable modeling of scientific and causal dynamics.

Retrieval-Augmented Generation (RAG) systems face challenges with complex, multihop questions, and agentic frameworks such as Search-R1 (Jin et al., 2025), which operates iteratively, have been proposed to address these complexities. However, such approaches can introduce inefficiencies, including repetitive retrieval of previously processed information and challenges in contextualizing retrieved results effectively within the current generation prompt. Such issues can lead to unnecessary retrieval turns, suboptimal reasoning, inaccurate answers, and increased token consumption. In this paper, we investigate test-time modifications to the Search-R1 pipeline to mitigate these identified shortcomings. Specifically, we explore the integration of two components and their combination: a contextualization module to better integrate relevant information from retrieved documents into reasoning, and a de-duplication module that replaces previously retrieved documents with the next most relevant ones. We evaluate our approaches using the HotpotQA (Yang et al., 2018) and the Natural Questions (Kwiatkowski et al., 2019) datasets, reporting the exact match (EM) score, an LLM-as-a-Judge assessment of answer correctness, and the average number of turns. Our best-performing variant, utilizing GPT-4.1-mini for contextualization, achieves a 5.6% increase in EM score and reduces the number of turns by 10.5% compared to the Search-R1 baseline, demonstrating improved answer accuracy and retrieval efficiency.

Diffusion and flow models achieve State-Of-The-Art (SOTA) generative performance, yet many practically important behaviors such as fine-grained prompt fidelity, compositional correctness, and text rendering are weakly specified by score or flow matching pretraining objectives. Reinforcement Learning (RL) fine-tuning with external, black-box rewards is a natural remedy, but diffusion RL is often brittle. Trajectory-based methods incur high memory cost and high-variance gradient estimates; forward-process approaches converge faster but can suffer from distribution drift, and hence reward hacking. In this work, we present Centered Reward Distillation (CRD), a diffusion RL framework derived from KL-regularized reward maximization built on forward-process-based fine-tuning. The key insight is that the intractable normalizing constant cancels under within-prompt centering, yielding a well-posed reward-matching objective. To enable reliable text-to-image fine-tuning, we introduce techniques that explicitly control distribution drift: (i) decoupling the sampler from the moving reference to prevent ratio-signal collapse, (ii) KL anchoring to a CFG-guided pretrained model to control long-run drift and align with the inference-time semantics of the pre-trained model, and (iii) reward-adaptive KL strength to accelerate early learning under large KL regularization while reducing late-stage exploitation of reward-model loopholes. Experiments on text-to-image post-training with GenEval and OCR rewards show that CRD achieves competitive SOTA reward optimization results with fast convergence and reduced reward hacking, as validated on unseen preference metrics.

Scientific question answering requires a retrieval system to solve two distinct problems: identifying which papers are relevant and locating the supporting evidence within those papers. Conventional retrieval-augmented generation typically addresses both through similarity search over fixed-length passages, flattening document structure and separating scientific claims from their methodological and argumentative context. We present VecTree-RAG, an agentic framework that assigns these tasks to complementary retrieval mechanisms. Vector search ranks compact document and section representations across the corpus, whereas reasoning-guided traversal of source-verified section trees localizes evidence within shortlisted papers. Full text is retained in a page store and exposed progressively only after structural localization. We evaluate VecTree-RAG on 300 QASPER questions, an open-access subset of 54 LitQA2 questions, and 49 multi-document MOSAIC questions. Compared with Dense RAG, reranked Dense RAG, RAPTOR, and Search-o1, VecTree-RAG obtained the highest observed answer score on all three benchmarks, reaching 0.800 LLM-judge correctness on QASPER, 0.925 accuracy on LitQA2, and a 0.547 composite score on MOSAIC. On QASPER, its evidence-page precision was 0.274, compared with 0.046--0.071 for the baselines. LitQA2 ablations further showed that the complete vector--tree architecture required fewer inference tokens than variants without tree navigation or corpus-level vector routing. These results indicate that vector retrieval narrows the corpus-level search space and tree navigation concentrates reading on structurally relevant evidence. Although multi-turn inference remains more expensive than single-call retrieval, VecTree-RAG provides a structure-aware and traceable architecture for scientific literature question answering.

Despite their vast capabilities, Large Language Models (LLMs) often struggle with generating reliable outputs, frequently producing high-confidence inaccuracies known as hallucinations. Addressing this challenge, our research introduces InternalInspector, a novel framework designed to enhance confidence estimation in LLMs by leveraging contrastive learning on internal states including attention states, feed-forward states, and activation states of all layers. Unlike existing methods that primarily focus on the final activation state, InternalInspector conducts a comprehensive analysis across all internal states of every layer to accurately identify both correct and incorrect prediction processes. By benchmarking InternalInspector against existing confidence estimation methods across various natural language understanding and generation tasks, including factual question answering, commonsense reasoning, and reading comprehension, InternalInspector achieves significantly higher accuracy in aligning the estimated confidence scores with the correctness of the LLM's predictions and lower calibration error. Furthermore, InternalInspector excels at HaluEval, a hallucination detection benchmark, outperforming other internal-based confidence estimation methods in this task.

Multimodal Retrieval-Augmented Generation (MRAG) enhances reasoning capabilities by integrating external knowledge. However, existing benchmarks primarily focus on simple image-text interactions, overlooking complex visual formats like charts that are prevalent in real-world applications. In this work, we introduce a novel task, Chart-based MRAG, to address this limitation. To semi-automatically generate high-quality evaluation samples, we propose CHARt-based document question-answering GEneration (CHARGE), a framework that produces evaluation data through structured keypoint extraction, crossmodal verification, and keypoint-based generation. By combining CHARGE with expert validation, we construct Chart-MRAG Bench, a comprehensive benchmark for chart-based MRAG evaluation, featuring 4,738 question-answering pairs across 8 domains from real-world documents. Our evaluation reveals three critical limitations in current approaches: (1) unified multimodal embedding retrieval methods struggles in chart-based scenarios, (2) even with ground-truth retrieval, state-of-the-art MLLMs achieve only 58.19% Correctness and 73.87% Coverage scores, and (3) MLLMs demonstrate consistent text-over-visual modality bias during Chart-based MRAG reasoning. The CHARGE and Chart-MRAG Bench are released at https://github.com/Nomothings/CHARGE.git.

Large Language Models are prompting us to view more NLP tasks from a generative perspective. At the same time, they offer a new way of accessing information, mainly through the RAG framework. While there have been notable improvements for the autoregressive models, overcoming hallucination in the generated answers remains a continuous problem. A standard solution is to use commercial LLMs, such as GPT4, to evaluate these algorithms. However, such frameworks are expensive and not very transparent. Therefore, we propose a study which demonstrates the interest of open-weight models for evaluating RAG hallucination. We develop a lightweight approach using smaller, quantized LLMs to provide an accessible and interpretable metric that gives continuous scores for the generated answer with respect to their correctness and faithfulness. This score allows us to question decisions' reliability and explore thresholds to develop a new AUC metric as an alternative to correlation with human judgment.

Reward models can significantly enhance the reasoning capabilities of large language models (LLMs), but they typically require extensive curated data and costly training. To mitigate these challenges, training-free approaches such as LLM-as-a-Judge leverage the intrinsic reasoning abilities of LLMs to evaluate responses, achieving promising results. Recent works have also indicated that model confidence can serve effectively as a reward metric, distinguishing between chain-of-thought (CoT) and non-CoT paths. However, the concept of using confidence as a reward has not been comprehensively studied. In this work, we systematically investigate Confidence-as-a-Reward (CRew), a simple yet powerful training-free method that utilizes token-level confidence in the model's final answers as a proxy for reward, especially suitable for close-ended tasks. Through extensive experiments on mathematical reasoning tasks, we demonstrate that CRew outperforms existing training-free reward approaches on the MATH500 and RewardMATH benchmarks, and even surpasses most trained reward models. We further identify a strong correlation between CRew scores and the actual reasoning performance of the model. Additionally, we find that CRew can effectively filter high-quality training data. Building upon these insights, we propose CRew-DPO, a training strategy that constructs preference data from confidence scores combined with correctness signals. Finetuning with CRew-DPO further enhances the model's judging capabilities and consistently outperforms existing self-training methods.

Routing has emerged as a promising strategy for balancing performance and cost in large language model (LLM) systems that combine lightweight models with powerful but expensive large models. Recent studies show that probe routing, which predicts the correctness of a small model using its hidden states, provides an effective solution in text-only LLMs. However, we observe that these probes degrade substantially when applied to multimodal LLMs (MLLMs). Through empirical analysis, we find that the presence of visual inputs weakens the separability of correctness signals in hidden states, making them harder to extract using standard probe designs. To address this challenge, we introduce two complementary approaches for improving probe routing in MLLMs. First, we propose the Attention Probe, which aggregates hidden states from the preceding layer based on attention scores to recover distributed correctness signals. Second, we present the KL-Regularized LoRA Probe (ReLope), which inserts a lightweight LoRA adapter and applies a KL regularizer to learn routing-aware representations. Comprehensive experiments show that our methods consistently outperform baselines, suggesting that improving the quality of hidden states is key to effective routing in MLLMs. Our code is available at https://github.com/Spinozaaa/ReLope.

We present Metal-Sci, a 10-task benchmark of scientific Apple Silicon Metal compute kernels spanning six optimization regimes (stencils, all-pairs in n-body problems, multi-field Boltzmann, neighbor-list molecular dynamics, multi-kernel PDE, FFT). Each task ships a CPU reference, a roofline-anchored fitness function, and a held-out generalization size. We pair the benchmark with a lightweight harness for automatic kernel search that runtime-compiles each candidate, scores it against the roofline across multiple sizes, and feeds structured compile and per-size correctness diagnostics back to a frozen LLM driving a (1{+}1) evolutionary loop. We report matched single-model sweeps of Claude Opus 4.7, Gemini 3.1 Pro, and GPT 5.5 on M1 Pro: in-distribution self-speedups span 1.00times to 10.7times. Beyond raw speedup, our central methodological claim is structural: the held-out gate scoring function Φ_T (evaluated once at end-of-run on a configuration the agent never sees during search) functions as a cheap mechanical oversight primitive on this automatic search loop, catching e.g. an Opus template <uint D> HMC win that returns wrong samples at unseen dimensions, and a GPT FFT3D best that wins in-distribution at 2.95times speedup but collapses to 0.23times on a 256^3 held-out cube, a silent regression that the in-distribution score alone cannot see. Code at https://github.com/vicgalle/metal-sci-kernels

In industrial procurement, an LLM answer is useful only if it survives a standards check: recommended material must match operating condition, every parameter must respect a regulated threshold, and no procedure may contradict a safety clause. Partial correctness can mask safety-critical contradictions that aggregate LLM benchmarks rarely capture. We introduce IndustryBench, a 2,049-item benchmark for industrial procurement QA in Chinese, grounded in Chinese national standards (GB/T) and structured industrial product records, organized by seven capability dimensions, ten industry categories, and panel-derived difficulty tiers, with item-aligned English, Russian, and Vietnamese renderings. Our construction pipeline rejects 70.3% of LLM-generated candidates at a search-based external-verification stage, calibrating how unreliable industrial QA remains after LLM-only filtering.Our evaluation decouples raw correctness, scored by a Qwen3-Max judge validated at κ_w = 0.798 against a domain expert, from a separate safety-violation (SV) check against source texts. Across 17 models in Chinese and an 8-model intersection over four languages, we find: (i) the best system reaches only 2.083 on the 0--3 rubric, leaving substantial headroom; (ii) Standards & Terminology is the most persistent capability weakness and survives item-aligned translation; (iii) extended reasoning lowers safety-adjusted scores for 12 of 13 models, primarily by introducing unsupported safety-critical details into longer final answers; and (iv) safety-violation rates reshuffle the leaderboard -- GPT-5.4 climbs from rank 6 to rank 3 after SV adjustment, while Kimi-k2.5-1T-A32B drops seven positions.Industrial LLM evaluation therefore requires source-grounded, safety-aware diagnosis rather than aggregate accuracy. We release IndustryBench with all prompts, scoring scripts, and dataset documentation.

Text- and image-conditioned video generation models have achieved strong visual fidelity and temporal coherence, but they often fail to generate motion governed by kinematic and geometric constraints. In these settings, object parts must remain rigid, maintain contact or coupling with neighboring components, and transfer motion consistently across connected parts. These requirements are especially explicit in articulated mechanical assemblies, where motion is constrained by rigid-link geometry, contact/coupling relations, and transmission through kinematic chains. A generated video may therefore appear plausible while violating the intended mechanism, such as rotating a part that should translate, deforming a rigid component, breaking coupling between parts, or failing to move downstream components. To evaluate this gap, We introduce MechVerse, a benchmark for mechanically consistent image-to-video generation. MechVerse contains 21,156 synthetic clips from 1,357 mechanical assemblies across 141 categories, organized into three tiers of increasing kinematic complexity: independent articulation, pairwise coupling, and densely coupled multi-part mechanisms. Each clip is paired with a structured prompt describing part identities, stationary supports, moving components, motion primitives, direction, speed/extent, and inter-part dependencies. We evaluate proprietary, open-source, and fine-tuned image-to-video models using standard video metrics, instruction-following scores, and human judgments of motion correctness and kinematic coupling. Results show that current models can preserve appearance and smoothness while failing to generate mechanically admissible motion, with errors increasing as coupling complexity grows. MechVerse provides a benchmark for measuring and improving mechanism-aware video generation from image and language inputs.

This paper presents BR-TaxQA-R, a novel dataset designed to support question answering with references in the context of Brazilian personal income tax law. The dataset contains 715 questions from the 2024 official Q\&A document published by Brazil's Internal Revenue Service, enriched with statutory norms and administrative rulings from the Conselho Administrativo de Recursos Fiscais (CARF). We implement a Retrieval-Augmented Generation (RAG) pipeline using OpenAI embeddings for searching and GPT-4o-mini for answer generation. We compare different text segmentation strategies and benchmark our system against commercial tools such as ChatGPT and Perplexity.ai using RAGAS-based metrics. Results show that our custom RAG pipeline outperforms commercial systems in Response Relevancy, indicating stronger alignment with user queries, while commercial models achieve higher scores in Factual Correctness and fluency. These findings highlight a trade-off between legally grounded generation and linguistic fluency. Crucially, we argue that human expert evaluation remains essential to ensure the legal validity of AI-generated answers in high-stakes domains such as taxation. BR-TaxQA-R is publicly available at https://huggingface.co/datasets/unicamp-dl/BR-TaxQA-R.

Benchmarks for evaluating large language models on register-transfer-level (RTL) hardware design have proliferated rapidly, yet none reports having applied mutation testing, an established hardware-verification technique for quantifying testbench quality, to ask whether its own testbenches are trustworthy. A testbench that never fails is not evidence of a correct design; it may simply never stimulate the logic that is actually broken. We introduce GateTruth, a mutation-testing engine and methodology for auditing RTL benchmark testbench rigor: inject a deterministic, seeded set of semantic mutants into a reference design and measure what fraction the testbench catches. We validate the methodology against our own 68-task, dual-track reference suite -- 60 specification-to-RTL generation tasks and 8 agentic-repair tasks, scored through a pinned, deterministic synthesis-to-timing flow with correctness enforced as a strict gate -- certifying that 46 of 60 Track A testbenches kill at least 95% of injected mutants under sequential, reproducible execution; we disclose why the other 14 do not, including a Goodhart effect on testbenches revised to pass this gate. We then point the same engine, unmodified, at RTLLM v2.0, a widely adopted external benchmark: of 46 auditable designs, 72% fall below the 95% floor our own suite is held to, and three score 0% outright. A comparable audit of NVIDIA's CVDP benchmark is structurally impossible: its public release withholds reference solutions, removing the golden RTL mutation testing requires. Auditing our own instrument also surfaced a second finding: an initially uniform 4096-token output cap silently truncated three of seven evaluated models, and re-running at 16,384 tokens moved one model from fifth place to first. We argue mutation-kill certification should become a standard reporting requirement for RTL-generation benchmarks generally.

Aligning multimodal large language models (MLLMs) with human preferences often relies on single-signal, model-based reward methods. Such monolithic rewards often lack confidence calibration across domain-specific tasks, fail to capture diverse aspects of human preferences, and require extensive data annotation and reward model training. In this work, we propose a hybrid reward modeling framework that integrates complementary reward paradigms: (i) model-based rewards, where a learned reward model predicts scalar or vector scores from synthetic and human feedback, and (ii) rule-based rewards, where domain-specific heuristics provide explicit correctness signals with confidence. Beyond accuracy, we further incorporate multi-aspect rewards to enforce instruction adherence and introduce a generalized length-penalty reward to stabilize training and improve performance. The proposed framework provides a flexible and effective approach to aligning MLLMs through reinforcement learning policy optimization. Our experiments show consistent improvements across different multimodal benchmarks when applying hybrid and multi-aspect reward modeling. Our best performing model in the 3B family achieves an overall average improvement of ~9.5% across general and math reasoning tasks. Focusing specifically on mathematical benchmarks, the model achieves a significant average improvement of ~16%, highlighting its effectiveness in mathematical reasoning and problem solving.

This paper introduces CoMuRoS (Collaborative Multi-Robot System), a generalizable hierarchical architecture for heterogeneous robot teams that unifies centralized deliberation with decentralized execution, and supports event-driven replanning. A Task Manager LLM interprets natural-language goals, classifies tasks, and allocates subtasks using static rules plus dynamic contexts (task, history, robot and task status, and events).Each robot runs a local LLM that composes executable Python code from primitive skills (ROS2 nodes, policies), while onboard perception (VLMs/image processing) continuously monitors events and classifies them into relevant or irrelevant to the task. Task failures or user intent changes trigger replanning, allowing robots to assist teammates, resume tasks, or request human help. Hardware studies demonstrate autonomous recovery from disruptive events, filtering of irrelevant distractions, and tightly coordinated transport with emergent human-robot cooperation (e.g., multirobot collaborative object recovery success rate: 9/10, coordinated transport: 8/8, human-assisted recovery: 5/5).Simulation studies show intention-aware replanning. A curated textual benchmark spanning 22 scenarios (3 tasks each, around 20 robots) evaluates task allocation, classification, IoU, executability, and correctness, with high average scores (e.g., correctness up to 0.91) across multiple LLMs, a separate replanning set (5 scenarios) achieves 1.0 correctness. Compared with prior LLM-based systems, CoMuRoS uniquely demonstrates runtime, event-driven replanning on physical robots, delivering robust, flexible multi-robot and human-robot collaboration.

Hallucination mitigation remains a persistent challenge for large language models (LLMs), even as model scales grow. Existing approaches often rely on external knowledge sources, such as structured databases or knowledge graphs, accessed through prompting or retrieval. However, prompt-based grounding is fragile and domain-sensitive, while symbolic knowledge integration incurs heavy retrieval and formatting costs. Motivated by knowledge graphs, we introduce Graph-Retrieved Adaptive Decoding (GRAD), a decoding-time method that grounds generation in corpus-derived evidence without retraining. GRAD constructs a sparse token transition graph by accumulating next-token logits across a small retrieved corpus in a single forward pass. During decoding, graph-retrieved logits are max-normalized and adaptively fused with model logits to favor high-evidence continuations while preserving fluency. Across three models and a range of question-answering benchmarks spanning intrinsic, extrinsic hallucination, and factuality tasks, GRAD consistently surpasses baselines, achieving up to 9.7% higher intrinsic accuracy, 8.6% lower hallucination rates, and 6.9% greater correctness compared to greedy decoding, while attaining the highest truth--informativeness product score among all methods. GRAD offers a lightweight, plug-and-play alternative to contrastive decoding and knowledge graph augmentation, demonstrating that statistical evidence from corpus-level token transitions can effectively steer generation toward more truthful and verifiable outputs.

Critique, as a natural language description for assessing the quality of model-generated content, has been proven to play an essential role in the training, evaluation, and refinement of Large Language Models (LLMs). However, there is a lack of principled understanding in evaluating the quality of the critique itself. In this paper, we pioneer the critique of critique, termed MetaCritique, which is a framework to evaluate the critique from two aspects, i.e., factuality as precision score and comprehensiveness as recall score. We calculate the harmonic mean of precision and recall as the overall rating called F1 score. To obtain a reliable evaluation outcome, we propose Atomic Information Units (AIUs), which describe the critique in a more fine-grained manner. MetaCritique takes each AIU into account and aggregates each AIU's judgment for the overall score. Moreover, given the evaluation process involves intricate reasoning, our MetaCritique provides a natural language rationale to support each judgment. We construct a meta-evaluation dataset containing 300 critiques (2653 AIUs) across four tasks (question answering, reasoning, entailment, and summarization), and we conduct a comparative study to demonstrate the feasibility and effectiveness. Experiments also show superior critique judged by MetaCritique leads to better refinement, indicating generative artificial intelligence indeed has the potential to be significantly advanced with our MetaCritique. We will release relevant code and meta-evaluation datasets at https://github.com/GAIR-NLP/MetaCritique.

Generating accurate and calibrated confidence estimates is critical for deploying LLMs in high-stakes or user-facing applications, and remains an open challenge. Prior research has often framed confidence as a problem of eliciting a model's "self-knowledge", i.e., the ability of an LLM to judge whether its own answers are correct; this approach implicitly assumes that there is some privileged information about the answer's correctness that is accessible to the model itself. However, our experiments reveal that an LLM attempting to predict the correctness of its own outputs generally performs no better than an unrelated LLM. Moreover, we hypothesize that a key factor in building a "Correctness Model" (CM) is exposure to a target model's historical predictions. We propose multiple methods to inject this historical correctness information, creating a Generalized Correctness Model (GCM). We first show that GCMs can be trained on the correctness data from many LLMs and learn patterns for correctness prediction applicable across datasets and models. We then use CMs as a lens for studying the source of correctness prediction ability and its generalization, systematically controlling their training data and finding that answer phrasing is a strong predictor for correctness. We further explore alternative methods of injecting history without training an LLM, finding that including history as in-context examples can help improve correctness prediction, and post-hoc calibration can provide complementary reductions in calibration error. We evaluate GCMs based on Qwen3-8B across 5 model families and the MMLU and TriviaQA datasets, as well as on a downstream selective prediction task, finding that reliable LLM confidence estimation is a generalizable and model-agnostic skill learned by systematically encoding correctness history rather than a model-specific skill reliant on self-introspection.

Software is used in critical applications in our day-to-day life and it is important to ensure its correctness. One popular approach to assess correctness is to evaluate software on tests. If a test fails, it indicates a fault in the software under test; if all tests pass correctly, one may assume that the software is correct. However, the reliability of these results depends on the test suite considered, and there is a risk of false negatives (i.e. software that passes all available tests but contains bugs because some cases are not tested). Therefore, it is important to consider error-inducing test cases when evaluating software. To support data-driven creation of such a test-suite, which is especially of interest for testing software synthesized from large language models, we curate a dataset (Codehacks) of programming problems together with corresponding error-inducing test cases (i.e., "hacks"). This dataset is collected from the wild, in particular, from the Codeforces online judge platform. The dataset comprises 288,617 hacks for 5,578 programming problems, each with a natural language description, as well as the source code for 2,196 submitted solutions to these problems that can be broken with their corresponding hacks. Keywords: competitive programming, language model, dataset

Unstructured clinical notes contain essential patient information but are challenging for physicians to search and interpret efficiently. Although large language models (LLMs) have shown promise in question answering (QA), most existing systems lack transparency, usability, and alignment with clinical workflows. This work introduces an interactive QA system that enables physicians to query clinical notes via text or voice and receive extractive answers highlighted directly in the note for traceability. The system was built using OpenAI models with zero-shot prompting and evaluated across multiple metrics, including exact string match, word overlap, SentenceTransformer similarity, and BERTScore. Results show that while exact match scores ranged from 47 to 62 percent, semantic similarity scores exceeded 87 percent, indicating strong contextual alignment even when wording varied. To assess usability, the system was also evaluated using simulated clinical personas. Seven diverse physician and nurse personas interacted with the system across scenario-based tasks and provided structured feedback. The evaluations highlighted strengths in intuitive design and answer accessibility, alongside opportunities for enhancing explanation clarity.

AI agents are changing the requirements for document parsing. What matters is semantic correctness: parsed output must preserve the structure and meaning needed for autonomous decisions, including correct table structure, precise chart data, semantically meaningful formatting, and visual grounding. Existing benchmarks do not fully capture this setting for enterprise automation, relying on narrow document distributions and text-similarity metrics that miss agent-critical failures. We introduce ParseBench, a benchmark of {sim}2{,}000 human-verified pages from enterprise documents spanning insurance, finance, and government, organized around five capability dimensions: tables, charts, content faithfulness, semantic formatting, and visual grounding. Across 14 methods spanning vision-language models, specialized document parsers, and LlamaParse, the benchmark reveals a fragmented capability landscape: no method is consistently strong across all five dimensions. LlamaParse Agentic achieves the highest overall score at \agenticoverall\%, and the benchmark highlights the remaining capability gaps across current systems. Dataset and evaluation code are available on https://huggingface.co/datasets/llamaindex/ParseBench{HuggingFace} and https://github.com/run-llama/ParseBench{GitHub}.

Existing approaches to building line-level Arabic handwritten-text-recognition (HTR) training data either rely on fully manual annotation, which does not scale, or on automatic OCR-to-reference alignment methods not yet extended to multi-script, two-zone (main-plus-margin) manuscript layouts with a provable correctness guarantee. We present RefLAM (Reference-grounded Line Annotation for Manuscripts), a pipeline converting manuscript page images and clean transcriptions into validated, line-level ground truth without sacrificing human oversight. RefLAM couples a deep-learning page-segmentation model with a multimodal large language model (MLLM) for structured OCR and a diacritic-agnostic fuzzy alignment engine that grounds each OCR line in a contiguous span of the reference text, with a character-level confidence score in [0,100]. A perfect score is provably equivalent to character-for-character identity of the normalised strings (the Confidence-100 rule), verified with no counterexample across the released corpus. A reviewer can thus trust a perfect score, confirming most lines at a glance rather than retyping them, so annotation becomes triaged, with attention concentrated on uncertain alignments. Across 7 fully page-validated books we measured a 75times throughput gain over manual annotation (3,000 vs. 40 lines/hr); applying the same guarantee to 7 further books, we retained 16,533 confidence-100 main-text lines within one week, excluding sub-100 lines rather than manually correcting them. Using RefLAM, we release AraMS-28k: 14 historical Arabic manuscript books, 3,043 pages, and 27,971 main-text and 629 margin-line annotations with bounding boxes, layout labels, and insertion anchors for 191 margin entries (30.4%). We also finetune Muharaf-pretrained baselines (including HATFormer) on AraMS-28k and report CER results confirming its practical utility for downstream HTR training.

We provide here a dataset for tasks related to natural language understanding and natural language inference. The dataset contains logical puzzles in natural language from three domains: comparing puzzles, knighs and knaves, and zebra puzzles. Each puzzle is associated with the entire set of atomic questions that can be generated based on the relations and individuals occurring in the text. For each question we provide the correct answer: entailment, contradiction or ambiguity. The answer's correctness is verified against theorem provers. Good puzzles have two properties: (i) each piece of information is necessary and (ii) no unnecessary information is provided. These properties make puzzles interesting candidates for machine comprehension tasks.

While there are numerous benchmarks comparing the performance of modern language models (LMs), end-task evaluations often conflate notions of *factual accuracy* ("truth") and *reasoning ability* ("rationality", or "honesty" in the sense of correctly reporting implications of beliefs). Our goal is a dataset that clearly distinguishes these two notions. Our approach is to leverage and extend a collection of human-annotated *entailment trees*, engineered to express both good and bad chains of reasoning, and using a mixture of true and false facts, in particular including counterfactual examples, to avoid belief bias (also known as the "content effect"). The resulting dataset, called BaRDa, contains 3000 entailments (1787 valid, 1213 invalid), using 6681 true and 2319 false statements. Testing on four GPT-series models, GPT3(curie)/GPT3(davinici)/3.5/4, we find factual accuracy (truth) scores of 74.1/80.6/82.6/87.1 and reasoning accuracy scores of 63.1/78.0/71.8/79.2. This shows the clear progression of models towards improved factual accuracy and entailment reasoning, and the dataset provides a new benchmark that more cleanly separates and quantifies these two notions.

Training large language models (LLMs) to write register-transfer level (RTL) requires large corpora of paired specifications and code, and such data is scarce enough that most public corpora are now synthesized. Synthesis provides scale but not correctness, and in two widely used RTL datasets only 24.4% and 53.5% of pairs pass generated functional tests. This raises the question of how much of such a corpus to keep and which part of it. Correctness alone is a poor answer. A pair that misbehaves in one corner case still shows valid syntax and interface conventions, and complex sequential designs are both harder to generate and harder to validate, so filtering by correctness leaves a corpus of short and simple modules. Correctness is also hard to obtain, since behavior leaves little trace on the surface in RTL, and validating an entire corpus only sorts pairs into passed and failed. We present RTLCurator, which learns a behavior-aware compatibility prior by contrasting each specification with implementations that fail simulation, and calibrates it to a new corpus using a small number of validated pairs. It then constructs the retained subset by balancing alignment, representation coverage, and RTL structural richness. On CodeV and RTLCoder, keeping 80% of the corpus this way improves on training with the full corpus across all reported metrics while validating only 10% of the pool, whereas ranking by the score alone falls below random selection and filtering the whole pool by simulation does no better.

Context: Code reviews are crucial for software quality. Recent AI advances have allowed large language models (LLMs) to review and fix code; now, there are tools that perform these reviews. However, their reliability and accuracy have not yet been systematically evaluated. Objective: This study compares different LLMs' performance in detecting code correctness and suggesting improvements. Method: We tested GPT4o and Gemini 2.0 Flash on 492 AI generated code blocks of varying correctness, along with 164 canonical code blocks from the HumanEval benchmark. To simulate the code review task objectively, we expected LLMs to assess code correctness and improve the code if needed. We ran experiments with different configurations and reported on the results. Results: With problem descriptions, GPT4o and Gemini 2.0 Flash correctly classified code correctness 68.50% and 63.89% of the time, respectively, and corrected the code 67.83% and 54.26% of the time for the 492 code blocks of varying correctness. Without problem descriptions, performance declined. The results for the 164 canonical code blocks differed, suggesting that performance depends on the type of code. Conclusion: LLM code reviews can help suggest improvements and assess correctness, but there is a risk of faulty outputs. We propose a process that involves humans, called the "Human in the loop LLM Code Review" to promote knowledge sharing while mitigating the risk of faulty outputs.

An abundance of datasets and availability of reliable evaluation metrics have resulted in strong progress in factoid question answering (QA). This progress, however, does not easily transfer to the task of long-form QA, where the goal is to answer questions that require in-depth explanations. The hurdles include (i) a lack of high-quality data, and (ii) the absence of a well-defined notion of the answer's quality. In this work, we address these problems by (i) releasing a novel dataset and a task that we call ASQA (Answer Summaries for Questions which are Ambiguous); and (ii) proposing a reliable metric for measuring performance on ASQA. Our task focuses on factoid questions that are ambiguous, that is, have different correct answers depending on interpretation. Answers to ambiguous questions should synthesize factual information from multiple sources into a long-form summary that resolves the ambiguity. In contrast to existing long-form QA tasks (such as ELI5), ASQA admits a clear notion of correctness: a user faced with a good summary should be able to answer different interpretations of the original ambiguous question. We use this notion of correctness to define an automated metric of performance for ASQA. Our analysis demonstrates an agreement between this metric and human judgments, and reveals a considerable gap between human performance and strong baselines.

We introduce a professionally translated extension of the TruthfulQA benchmark designed to evaluate truthfulness in Basque, Catalan, Galician, and Spanish. Truthfulness evaluations of large language models (LLMs) have primarily been conducted in English. However, the ability of LLMs to maintain truthfulness across languages remains under-explored. Our study evaluates 12 state-of-the-art open LLMs, comparing base and instruction-tuned models using human evaluation, multiple-choice metrics, and LLM-as-a-Judge scoring. Our findings reveal that, while LLMs perform best in English and worst in Basque (the lowest-resourced language), overall truthfulness discrepancies across languages are smaller than anticipated. Furthermore, we show that LLM-as-a-Judge correlates more closely with human judgments than multiple-choice metrics, and that informativeness plays a critical role in truthfulness assessment. Our results also indicate that machine translation provides a viable approach for extending truthfulness benchmarks to additional languages, offering a scalable alternative to professional translation. Finally, we observe that universal knowledge questions are better handled across languages than context- and time-dependent ones, highlighting the need for truthfulness evaluations that account for cultural and temporal variability. Dataset and code are publicly available under open licenses.

Self-correction of large language models (LLMs) emerges as a critical component for enhancing their reasoning performance. Although various self-correction methods have been proposed, a comprehensive evaluation of these methods remains largely unexplored, and the question of whether LLMs can truly correct themselves is a matter of significant interest and concern. In this study, we introduce CorrectBench, a benchmark developed to evaluate the effectiveness of self-correction strategies, including intrinsic, external, and fine-tuned approaches, across three tasks: commonsense reasoning, mathematical reasoning, and code generation. Our findings reveal that: 1) Self-correction methods can improve accuracy, especially for complex reasoning tasks; 2) Mixing different self-correction strategies yields further improvements, though it reduces efficiency; 3) Reasoning LLMs (e.g., DeepSeek-R1) have limited optimization under additional self-correction methods and have high time costs. Interestingly, a comparatively simple chain-of-thought (CoT) baseline demonstrates competitive accuracy and efficiency. These results underscore the potential of self-correction to enhance LLM's reasoning performance while highlighting the ongoing challenge of improving their efficiency. Consequently, we advocate for further research focused on optimizing the balance between reasoning capabilities and operational efficiency. Project Page: https://correctbench.github.io/

Humanity's Last Exam (HLE) has become a widely used benchmark for evaluating frontier large language models on challenging, multi-domain questions. However, community-led analyses have raised concerns that HLE contains a non-trivial number of noisy items, which can bias evaluation results and distort cross-model comparisons. To address this challenge, we introduce HLE-Verified, a verified and revised version of HLE with a transparent verification protocol and fine-grained error taxonomy. Our construction follows a two-stage validation-and-repair workflow resulting in a certified benchmark. In Stage I, each item undergoes binary validation of the problem and final answer through domain-expert review and model-based cross-checks, yielding 641 verified items. In Stage II, flawed but fixable items are revised under strict constraints preserving the original evaluation intent, through dual independent expert repairs, model-assisted auditing, and final adjudication, resulting in 1,170 revised-and-certified items. The remaining 689 items are released as a documented uncertain set with explicit uncertainty sources and expertise tags for future refinement. We evaluate seven state-of-the-art language models on HLE and HLE-Verified, observing an average absolute accuracy gain of 7--10 percentage points on HLE-Verified. The improvement is particularly pronounced on items where the original problem statement and/or reference answer is erroneous, with gains of 30--40 percentage points. Our analyses further reveal a strong association between model confidence and the presence of errors in the problem statement or reference answer, supporting the effectiveness of our revisions. Overall, HLE-Verified improves HLE-style evaluations by reducing annotation noise and enabling more faithful measurement of model capabilities. Data is available at: https://github.com/SKYLENAGE-AI/HLE-Verified

Machine reading comprehension (MRC) is a sub-field in natural language processing that aims to assist computers understand unstructured texts and then answer questions related to them. In practice, the conversation is an essential way to communicate and transfer information. To help machines understand conversation texts, we present UIT-ViCoQA, a new corpus for conversational machine reading comprehension in the Vietnamese language. This corpus consists of 10,000 questions with answers over 2,000 conversations about health news articles. Then, we evaluate several baseline approaches for conversational machine comprehension on the UIT-ViCoQA corpus. The best model obtains an F1 score of 45.27%, which is 30.91 points behind human performance (76.18%), indicating that there is ample room for improvement. Our dataset is available at our website: http://nlp.uit.edu.vn/datasets/ for research purposes.

The reliance on translated or adapted datasets from English or multilingual resources introduces challenges regarding linguistic and cultural suitability. This study addresses the need for robust and culturally appropriate benchmarks by evaluating the quality of 17 commonly used Turkish benchmark datasets. Using a comprehensive framework that assesses six criteria, both human and LLM-judge annotators provide detailed evaluations to identify dataset strengths and shortcomings. Our results reveal that 70% of the benchmark datasets fail to meet our heuristic quality standards. The correctness of the usage of technical terms is the strongest criterion, but 85% of the criteria are not satisfied in the examined datasets. Although LLM judges demonstrate potential, they are less effective than human annotators, particularly in understanding cultural common sense knowledge and interpreting fluent, unambiguous text. GPT-4o has stronger labeling capabilities for grammatical and technical tasks, while Llama3.3-70B excels at correctness and cultural knowledge evaluation. Our findings emphasize the urgent need for more rigorous quality control in creating and adapting datasets for low-resource languages.

Multi-modal generative document parsing systems challenge traditional evaluation: unlike deterministic OCR or layout models, they often produce semantically correct yet structurally divergent outputs. Conventional metrics-CER, WER, IoU, or TEDS-misclassify such diversity as error, penalizing valid interpretations and obscuring system behavior. We introduce SCORE (Structural and COntent Robust Evaluation), an interpretation-agnostic framework that integrates (i) adjusted edit distance for robust content fidelity, (ii) token-level diagnostics to distinguish hallucinations from omissions, (iii) table evaluation with spatial tolerance and semantic alignment, and (iv) hierarchy-aware consistency checks. Together, these dimensions enable evaluation that embraces representational diversity while enforcing semantic rigor. Across 1,114 pages spanning a holistic benchmark and a field dataset, SCORE consistently revealed cross-dataset performance patterns missed by standard metrics. In 2-5% of pages with ambiguous table structures, traditional metrics penalized systems by 12-25% on average, leading to distorted rankings. SCORE corrected these cases, recovering equivalence between alternative but valid interpretations. Moreover, by normalizing generative outputs into a format-agnostic representation, SCORE reproduces traditional scores (e.g., table F1 up to 0.93) without requiring object-detection pipelines, demonstrating that generative parsing alone suffices for comprehensive evaluation. By exposing how interpretive diversity impacts evaluation outcomes and providing multi-dimensional, interpretable diagnostics, SCORE establishes foundational principles for semantically grounded, fair, and practical benchmarking of modern document parsing systems.

Passage ranking involves two stages: passage retrieval and passage re-ranking, which are important and challenging topics for both academics and industries in the area of Information Retrieval (IR). However, the commonly-used datasets for passage ranking usually focus on the English language. For non-English scenarios, such as Chinese, the existing datasets are limited in terms of data scale, fine-grained relevance annotation and false negative issues. To address this problem, we introduce T2Ranking, a large-scale Chinese benchmark for passage ranking. T2Ranking comprises more than 300K queries and over 2M unique passages from real-world search engines. Expert annotators are recruited to provide 4-level graded relevance scores (fine-grained) for query-passage pairs instead of binary relevance judgments (coarse-grained). To ease the false negative issues, more passages with higher diversities are considered when performing relevance annotations, especially in the test set, to ensure a more accurate evaluation. Apart from the textual query and passage data, other auxiliary resources are also provided, such as query types and XML files of documents which passages are generated from, to facilitate further studies. To evaluate the dataset, commonly used ranking models are implemented and tested on T2Ranking as baselines. The experimental results show that T2Ranking is challenging and there is still scope for improvement. The full data and all codes are available at https://github.com/THUIR/T2Ranking/

We introduce SimpleQA Verified, a 1,000-prompt benchmark for evaluating Large Language Model (LLM) short-form factuality based on OpenAI's SimpleQA. It addresses critical limitations in OpenAI's benchmark, including noisy and incorrect labels, topical biases, and question redundancy. SimpleQA Verified was created through a rigorous multi-stage filtering process involving de-duplication, topic balancing, and source reconciliation to produce a more reliable and challenging evaluation set, alongside improvements in the autorater prompt. On this new benchmark, Gemini 2.5 Pro achieves a state-of-the-art F1-score of 55.6, outperforming other frontier models, including GPT-5. This work provides the research community with a higher-fidelity tool to track genuine progress in parametric model factuality and to mitigate hallucinations. The benchmark dataset, evaluation code, and leaderboard are available at: https://www.kaggle.com/benchmarks/deepmind/simpleqa-verified.

Large language models (LLMs) have experienced notable advancements in generating coherent and contextually relevant responses. However, hallucinations - incorrect or unfounded claims - are still prevalent, prompting the creation of automated metrics to detect these in LLM outputs. Our contributions include: introducing ChainPoll, an innovative hallucination detection method that excels compared to its counterparts, and unveiling RealHall, a refined collection of benchmark datasets to assess hallucination detection metrics from recent studies. While creating RealHall, we assessed tasks and datasets from previous hallucination detection studies and observed that many are not suitable for the potent LLMs currently in use. Overcoming this, we opted for four datasets challenging for modern LLMs and pertinent to real-world scenarios. Using RealHall, we conducted a comprehensive comparison of ChainPoll with numerous hallucination metrics from recent studies. Our findings indicate that ChainPoll outperforms in all RealHall benchmarks, achieving an overall AUROC of 0.781. This surpasses the next best theoretical method by 11% and exceeds industry standards by over 23%. Additionally, ChainPoll is cost-effective and offers greater transparency than other metrics. We introduce two novel metrics to assess LLM hallucinations: Adherence and Correctness. Adherence is relevant to Retrieval Augmented Generation workflows, evaluating an LLM's analytical capabilities within given documents and contexts. In contrast, Correctness identifies logical and reasoning errors.

Through reinforcement learning with verifiable rewards (RLVR), large language models have achieved substantial progress in domains with easily verifiable outcomes, such as mathematics and coding. However, when applied to more complex tasks like open-domain question answering, RLVR faces significant challenges due to the difficulty of verifying correctness. The nuanced and ambiguous nature of real-world knowledge makes it difficult to reliably evaluate correctness in these settings, necessitating further abilities that extend beyond mere logical consistency to encompass an understanding and assessment of both external and internal knowledge. Recent work has primarily focused on improving faithfulness, defined as semantic alignment with supporting documents, which can cause models to rely excessively on external sources and diminish their capacity for critical assessment. To address this, we propose the Thinking-supervised Reward Model (TRM), which incorporates sentence-level thinking supervision to endow reward models with critical thinking abilities. Given a query, answer, and supporting documents, TRM first assesses the faithfulness of each answer sentence to the supporting documents, and then applies a reasoning step to evaluate sentence-level correctness. By structuring reward modeling as a sequence of faithfulness, reasoning, and correctness evaluations, TRM encourages models to critically assess and leverage both external and internal knowledge. Experiments on reward signals demonstrate that TRM substantially improves the identification of incorrect sentences, and incorporating TRM into policy optimization leads to significant gains in both answer correctness and usefulness.

A large share of day-to-day work in professional domains happens inside PDF files: benefits packets, leases, datasheets, clinical guidelines, construction plans. Benchmarks for document AI have generally measured the required capabilities in isolation: OCR, layout analysis, chart reasoning, table QA, document VQA. A high score on any one of them does not necessarily reveal whether a model can answer a realistic question that someone in the field would actually ask about a specific PDF. GDP_pdf is a benchmark built to measure this directly. It consists of question-document pairs authored by working professionals in ten fields, and a candidate question was kept only when at least two frontier multimodal models failed it in a way that mattered: a wrong answer, missed decisive evidence, or a fabricated claim, rather than a superficial difference such as style. Each item comes with a rubric of atomic criteria, so we can report a graded rubric score as well as a strict task-level pass rate, and each item is tagged against a taxonomy of eleven capabilities in three tiers, spanning text extraction and grounding, table and chart comprehension, cross-referencing, spatial reasoning, and abstention on unsupported queries. We report results for seventeen frontier models on the 100-item benchmark: the best model passes only 30.7% of the items and the worst passes 2%. Most errors trace back to a small set of recurring loss patterns: misaligned tables, misread charts, skipped footnotes and exclusions, miscounted floor-plan symbols, scan noise, and amendments that supersede earlier text.