Daily Papers

Deep research agents increasingly automate complex information-seeking tasks, producing evidence-grounded reports via multi-step reasoning, tool use, and synthesis. Their growing role demands scalable, reliable evaluation, positioning LLM-as-judge as a supervision paradigm for assessing factual accuracy, evidence use, and reasoning quality. Yet the reliability of these judges for deep research agents remains poorly understood, posing a critical meta-evaluation problem: before deploying LLM judges to supervise research agents, we must first evaluate the judges themselves. Existing meta-evaluations fall short in two ways: (1) reliance on coarse, subjective human-preference agreement; (2) focus on instruction-following or verifiable tasks, leaving open-ended agent executions unexplored. To address these gaps, we introduce REFLECT (REliable Fine-grained LLM judge Evaluation via Controlled inTervention), a meta-evaluation benchmark targeting fine-grained failure detection in agentic environments. REFLECT defines a detailed taxonomy of process- and outcome-level failure modes, instantiated by performing controlled and localized interventions on quality-screened agent execution traces. This yields verifiable, comprehensive, and fine-grained instances for validating the judge models. Our experiments show that current LLM judges remain unreliable: even the best-performing models achieve overall accuracies below 55% across reasoning, tool-use, and report-quality failures, with especially poor performance on evidence verification. Together, our taxonomy and findings expose systematic judge limitations, reveal tradeoffs in cost and reliability, and offer actionable guidance for building more reliable evaluation pipelines for deep research agents.

Benchmarks for scientific artificial intelligence are mostly written to be scored: multiple-choice questions, curated agent tasks with reference solutions, or simulators with a known generative structure. Real scientific requests arrive differently. They are underspecified, they carry attachments, and lack ground truth. We report K-Bench 01, an evaluation built from first-turn requests sampled from live user traffic on K-Dense Web and run end to end by nine frontier models in identical sandboxes, yielding 1,602 completed agent runs. Three blinded language-model judges scored every run against an eight-dimension rubric. On a rubric whose 8-anchor instructs judges that a domain scientist would accept the work with minor edits, no model clears the line under all three judges. gpt-5.6-sol has the highest pooled mean, 8.04, but its 95% interval [7.80, 8.23] spans the threshold, and two of the three judges rank claude-opus-5 first instead. We therefore report the ordering of systems as the reproducible quantity, the absolute level as an attribute of the instrument, and the top of the table as unresolved. Across all 39,934 scored judgments -- the eight dimension scores plus a holistic overall for each assessment, excluding not-applicable cells -- 47.6% fall below the 8-point threshold. Difficulty is not uniform across the rubric: scientific accuracy averages 6.22 against 7.33 for communication, on identical denominators and in the same direction within every one of the nine models. The single leading failure tag is overclaiming, on 31.4% of assessments. We argue that the informative quantity for scientific agents is not a leaderboard position but the joint distribution of what was delivered, what was claimed, and what artifacts were produced.

As interactive LLM-based applications are created and refined, model developers need to evaluate the quality of generated text along many possible axes. For simpler systems, human evaluation may be practical, but in complicated systems like conversational chatbots, the amount of generated text can overwhelm human annotation resources. Model developers have begun to rely heavily on auto-evaluation, where LLMs are also used to judge generation quality. However, existing LLM-as-a-judge benchmarks largely focus on simple Q\&A tasks that do not match the complexity of multi-turn conversations. We introduce RankJudge, a benchmark generator for evaluating LLM-as-a-judge on multi-turn conversations grounded in reference documents. RankJudge creates pairs of conversations where one conversation has a single flaw injected into one turn. This construction allows paired conversations to be labeled unambiguously as better or worse, and precisely isolates failure categories to individual turns, enabling a strict joint correctness criterion for judging. We implement RankJudge across the domains of machine learning, biomedicine, and finance, evaluate 21 frontier LLM judges, and rank those judges via the Bradley-Terry model. Our formulation also allows ranking each conversation pair with difficulty ratings, which we use to dynamically curate the evaluation slice to reduce label noise, as confirmed via human annotation. We find that judge rankings are stable under partial observability, coarser correctness criteria, and an alternative random-walk rating algorithm.

Reasoning LLMs-as-Judges, which can benefit from inference-time scaling, provide a promising path for extending the success of reasoning models to non-verifiable domains where the output correctness/quality cannot be directly checked. However, while reasoning judges have shown better performance on static evaluation benchmarks, their effectiveness in actual policy training has not been systematically examined. Therefore, we conduct a rigorous study to investigate the actual impact of non-reasoning and reasoning judges in reinforcement-learning-based LLM alignment. Our controlled synthetic setting, where a "gold-standard" judge (gpt-oss-120b) provides preference annotations to train smaller judges, reveals key differences between non-reasoning and reasoning judges: non-reasoning judges lead to reward hacking easily, while reasoning judges can lead to policies that achieve strong performance when evaluated by the gold-standard judge. Interestingly, we find that the reasoning-judge-trained policies achieve such strong performance by learning to generate highly effective adversarial outputs that can also score well on popular benchmarks such as Arena-Hard by deceiving other LLM-judges. Combined with our further analysis, our study highlights both important findings and room for improvements for applying (reasoning) LLM-judges in non-verifiable LLM post-training.

Aligning Text-to-Image (T2I) generation models with human preferences increasingly relies on image reward models that score or rank generated images according to prompt alignment and perceptual quality. Existing reward models are commonly trained as Bradley-Terry (BT) preference models on large-scale human preference corpora, making them costly to train, difficult to adapt, and opaque in their evaluation criteria. Meanwhile, Vision-Language Model (VLM) judges can provide more fine-grained assessments through textual rubrics, but their manually designed or heuristically generated scoring rules may fail to reliably reflect human preferences. In this paper, we propose AutoRubric-T2I, the first rubric learning framework in T2I that automatically synthesizes and selects explicit rubrics for guiding VLM judges. AutoRubric-T2I first synthesizes reasoning traces from preference pairs into candidate rubrics, then uses a VLM judge to score paired images under each rubric, producing pairwise rubric-score differences for preference learning. To remove noisy and redundant rules, we further employ a ell_1-Regularized Logistic Regression Refiner, which selects the Top-N most discriminative rubrics. Extensive evaluations show that AutoRubric-T2I produces high-quality, interpretable reward signals using less than 0.01% of the annotated preference data, substantially reducing the need for large-scale reward-model training. On image reward benchmarks such as MMRB2, AutoRubric-T2I outperforms strong reward model baselines. We further validate AutoRubric-T2I as an RL reward on downstream T2I tasks, including TIIF and UniGenBench++, where it improves generation quality over scalar reward models using the Flow-GRPO pipeline on diffusion models.

While Large Language Models (LLMs) are increasingly adopted as automated judges for evaluating generated text, their outputs are often costly, and highly sensitive to prompt design, language, and aggregation strategies, severely, which limits reproducibility. To address these challenges, we propose \textit{OmniScore}, a family of complementary, deterministic learned metrics developed using small size (<1B) parameter models. OmniScore approximates LLM-judge behavior while preserving the low latency and consistency of traditional model-based scoring. We trained the models large-scale synthetic supervision (sim564k instances, in 107 languages) and evaluated using 8,617 manually annotated instances. The OmniScore family supports reliable, multi-dimensional scores across a variety of settings, including reference-based, source-grounded, and hybrid evaluations. We evaluate these models across question answering (QA), translation, and summarization in 6 languages. Our results demonstrate that lightweight, deterministic learned metrics provide a highly practical and scalable alternative to frontier LLMs. Our models and datasets can be found at https://huggingface.co/collections/QCRI/omniscore

We introduce MetroLLM-Bench, a 955-case benchmark for testing language models as the policy layer of a transit kiosk. It covers six real metro systems, ranging from 37 to 414 stations, and eleven categories that include routing, fare calculation, disruptions, accessibility, and adversarial input. In each case, the model must call structured tools and submit a machine-renderable terminal state containing an outcome, a per-ticket fare quote when applicable, and a kiosk action. Fourteen deterministic scoring components form Tier 1; eight semantic-quality components form Tier 2, six of which use a language-model judge. We report Tier 1 and the combined score of both tiers. A stratified 75/25 split reserves 717 cases for training-data generation and 238 for held-out evaluation. We evaluate twenty-six models from six vendors, of which twenty-three are ranked. On the held-out partition, a 4B Qwen 3.5 student trained through parameter-efficient fine-tuning (PEFT) exceeds both GPT-5.6 tiers on Tier 1 (91.3 against 90.6 and 90.0) and matches GPT-5.4 full at maximum reasoning effort (91.4), with a 2.6 GB Q4_K_M footprint. Larger 9B and 27B students provide no further Tier 1 improvement over the 4B student at this training scale. Across the four Qwen sizes, the PEFT gain over the corresponding base model decreases from +7.03 points at 2B (three training seeds) to -0.91 at 27B; every seed shows the same direction at every size. A deterministic rule-based baseline reaches 84.6 on Tier 1, with the remaining language-model advantage concentrated in policy adaptation, compound scenarios, accessibility, and temporal reasoning. Muse Glimmer 30B leads the composite ranking, and serving configuration alone moves the Qwen 3.5-to-3.8 comparison by 2.7 Tier 1 points. The benchmark, harness, reproduction guide, and fine-tuned students are released at https://github.com/continker/metrollm-bench.

Matching submissions with suitable reviewers at scale is a growing challenge for major venues, yet existing approaches either rely on coarse proxy signals that conflate general relatedness with true suitability, or require expensive human annotations that are difficult to scale for training. We propose MERIT, a two-stage framework that bridges this gap by converting criterion-level expertise matching into scalable suitability supervision. In the first stage, we train a reviewer assessor via reinforcement learning to identify the expertise dimensions a paper requires, match them against the reviewer's prior work, and produce a suitability decision, with rewards provided by an LLM judge guided by paper-specific expertise rubrics. In the second stage, we distill the assessor's predictions into an embedding-based retriever for efficient large-scale assignment. Experiments show that our 4B reviewer assessor outperforms larger general-purpose LLMs on suitability classification, and the resulting retriever achieves state-of-the-art performance across LR-Bench and the CMU Gold dataset. Our code is available at https://github.com/Luli3220/MERIT.

Most evaluations of large language models focus on standard tasks such as factual question answering or short summarization. This research expands that scope in two directions: first, by comparing two retrieval strategies, Graph RAG, structured knowledge-graph based, and Advanced RAG, hybrid keyword-semantic search, for QA; and second, by evaluating whether LLMs can generate high quality non-traditional academic outputs, specifically slide decks and podcast scripts. We implemented a prototype combining Meta's LLaMA 3 70B open weight and OpenAI's GPT 4o mini API based. QA performance was evaluated using both human ratings across eleven quality dimensions and large language model judges for scalable cross validation. GPT 4o mini with Advanced RAG produced the most accurate responses. Graph RAG offered limited improvements and led to more hallucinations, partly due to its structural complexity and manual setup. Slide and podcast generation was tested with document grounded retrieval. GPT 4o mini again performed best, though LLaMA 3 showed promise in narrative coherence. Human reviewers were crucial for detecting layout and stylistic flaws, highlighting the need for combined human LLM evaluation in assessing emerging academic outputs.

Generative image editing models struggle with structured statistical charts when data modifications require geometric synchronization. We formalize this task as Visuo-Logical Cascading Editing (VLCE). However, existing methods remain confined to localized text substitutions and struggle with dependency-aware cascading updates. To systematically evaluate this capability, we introduce ChartSync, an expert-validated benchmark constructed via a programmatic rendering pipeline that guarantees deterministic visuo-logical coupling for the ground truth. ChartSync comprises 870 triplets across 9 chart categories and 4 task types, including 235 geometry-coupled VLCE instances that specifically test cascading text-to-geometry synchronization. We further evaluate these instances via a two-tier framework combining objective visual metrics with a vision-language model judge paradigm to assess low-level fidelity alongside multimodal comprehension and reasoning. Evaluating 14 image editing models and one code-mediated pipeline reveals a nuanced capability gap: most open-source models suffer severe drops in geometric synchronization, while only two frontier proprietary models show emerging VLCE capability, with their residual errors mainly involving semantic isolation and background corruption. Our detailed error analysis deconstructs these failure paradigms to identify core meta-abilities for guiding future multimodal architectures. The ChartSync dataset and code are publicly released at https://github.com/kaka-yjk/ChartSyncCodebase.

Evaluating retrieval-augmented generation (RAG) systems traditionally relies on hand annotations for input queries, passages to retrieve, and responses to generate. We introduce ARES, an Automated RAG Evaluation System, for evaluating RAG systems along the dimensions of context relevance, answer faithfulness, and answer relevance. Using synthetic training data, ARES finetunes lightweight LM judges to assess the quality of individual RAG components. To mitigate potential prediction errors, ARES utilizes a small set of human-annotated datapoints for prediction-powered inference (PPI). Across six different knowledge-intensive tasks in KILT and SuperGLUE, ARES accurately evaluates RAG systems while using a few hundred human annotations during evaluation. Furthermore, ARES judges remain effective across domain shifts, proving accurate even after changing the type of queries and/or documents used in the evaluated RAG systems. We make our datasets and code for replication and deployment available at https://github.com/stanford-futuredata/ARES.

The Retrieval-Augmented Generation (RAG) systems' performance on Thai legal question answering is still limited, especially for questions requiring extensive, complex legal reasoning. To address these limitations, we introduce an approach aligning LLMs toward improved law citation accuracy and better response quality using Group-Relative Policy Optimization (GRPO). Our approach leverages BGE-M3 embeddings as a cost-efficient semantic-similarity reward, significantly reducing computational expenses up to 2.5x compared to large language model judges. Experiments on the NitiBench benchmark demonstrate substantial improvements: GRPO achieves up to 90% citation-F1 gains from the base model and a 31% increase in joint quality metrics over instruction tuning. Crucially, our method shows enhanced robustness on complex legal reasoning tasks compared to instruction tuning, providing an effective and resource-efficient solution for enhancing Thai legal LLMs.

Current preference learning methods achieve high accuracy on standard benchmarks but exhibit significant performance degradation when objective quality signals are removed. We introduce WritingPreferenceBench, a dataset of 1,800 human-annotated preference pairs (1,200 English, 600 Chinese) across 8 creative writing genres, where responses are matched for objective correctness, factual accuracy, and length. On this benchmark, sequence-based reward models--the standard architecture for RLHF--achieve only 52.7% mean accuracy, while zero-shot language model judges perform at 53.9%. In contrast, generative reward models that produce explicit reasoning chains achieve 81.8% accuracy. We observe high within-model variance across genres: individual models range from 18.2% to 81.8% accuracy across different writing categories, with standard deviations averaging 10.1%. This variance persists regardless of model scale, with 27B parameter models showing no consistent improvement over 8B variants. Our results suggest that current RLHF methods primarily learn to detect objective errors rather than capture subjective quality preferences (e.g., creativity, stylistic flair, and emotional resonance), and that successful preference modeling may require intermediate reasoning representations rather than direct classification.

As AI-assisted video creation becomes increasingly practical, instruction-guided video editing has become essential for refining generated or captured footage to meet professional requirements. Yet the field still lacks both a large-scale human-annotated dataset with complete editing examples and a standardized evaluator for comparing editing systems. Existing resources are limited by small scale, missing edited outputs, or the absence of human quality labels, while current evaluation often relies on expensive manual inspection or generic vision-language model judges that are not specialized for editing quality. We introduce VEFX-Dataset, a human-annotated dataset containing 5,049 video editing examples across 9 major editing categories and 32 subcategories, each labeled along three decoupled dimensions: Instruction Following, Rendering Quality, and Edit Exclusivity. Building on VEFX-Dataset, we propose VEFX-Reward, a reward model designed specifically for video editing quality assessment. VEFX-Reward jointly processes the source video, the editing instruction, and the edited video, and predicts per-dimension quality scores via ordinal regression. We further release VEFX-Bench, a benchmark of 300 curated video-prompt pairs for standardized comparison of editing systems. Experiments show that VEFX-Reward aligns more strongly with human judgments than generic VLM judges and prior reward models on both standard IQA/VQA metrics and group-wise preference evaluation. Using VEFX-Reward as an evaluator, we benchmark representative commercial and open-source video editing systems, revealing a persistent gap between visual plausibility, instruction following, and edit locality in current models.

Generative world models are increasingly used for video generation, where learned simulators are expected to capture the physical rules that govern real-world dynamics. However, evaluating whether generated videos actually follow these rules remains challenging. Existing physics-focused video benchmarks have made important progress, but they still face three key challenges, including the coarse evaluation frameworks that hide law-specific failures, response biases and fatigue that undermine the validity of annotation judgments, and automated evaluators that are insufficiently physics-aware or difficult to audit. To address those challenges, we introduce PhyGround, a criteria-grounded benchmark for evaluating physical reasoning in video generation. The benchmark contains 250 curated prompts, each augmented with an expected physical outcome, and a taxonomy of 13 physical laws across solid-body mechanics, fluid dynamics, and optics. Each law is operationalized through observable sub-questions to enable per-law diagnostics. We evaluate eight modern video generation models through a large-scale, quality-controlled human study, grounded on social science lab experiment design. A total of 459 annotators provided 5,796 complete annotations and over 37.4K fine-grained labels; after quality control, the retained annotations exhibited high split-half model-ranking correlations (Spearman's rho > 0.90). To support reproducible automated evaluation, we release PhyJudge-9B, an open physics-specialized VLM judge. PhyJudge-9B achieves substantially lower aggregate relative bias than Gemini-3.1-Pro (3.3% vs. 16.6%). We release prompts, human annotations, model checkpoints, and evaluation code on the project page https://phyground.github.io/.

Agentic red-teaming benchmarks report whether an injected agent was compromised as a single bit: the attack succeeded, or it did not. We argue that this binary attack-success rate discards the information a defender most needs, namely how harmful the resulting action was. We introduce an action-graded harm rubric that scores an agent's tool-call trajectory on a seven-level ordinal scale (L0 to L6) according to whether the executed action was reversible, whether it crossed scope to reach another party, and whether it expanded privilege. We compute the scale two ways: a deterministic oracle that reads the trajectory and the attacker's stated goal, and a panel of three frontier language-model judges that read a tag-free account of the same trajectory. Across four victim models and two defenses on the AgentDojo workspace suite, severity grading exposes three cases the binary metric hides, including a defense that reports a zero attack-success rate while still permitting an externally visible cross-scope leak through an unfiltered tool. The judge panel reproduces the oracle with high ordinal agreement (Krippendorff's alpha = 0.91) but shares systematic blind spots that we characterize, most notably a failure to recognize escalation chains. Unlike prior work that provides harm taxonomies, harmful-task completion tests, execution-level safety benchmarks, or severity-aware simulation, our contribution is a reusable, trace-grounded severity instrument applied to the actual actions recorded in existing red-team logs. All code, prompts, and per-episode logs are released.

Effective judges of Vision-Language Models (VLMs) are crucial for model development. Current methods for training VLM judges mainly rely on large-scale human preference annotations. However, such an approach is costly, and the annotations easily become obsolete as models rapidly improve. In this work, we present a framework to self-train a VLM judge model without any human preference annotations, using only self-synthesized data. Our method is iterative and has three stages: (1) generate diverse multimodal instruction-response pairs at varying quality levels, (2) generate reasoning traces and judgments for each pair, removing the ones that do not match our expected quality levels, and (3) training on correct judge answers and their reasoning traces. We evaluate the resulting judge on Multimodal RewardBench and VL-RewardBench across domains: correctness, preference, reasoning, safety, and visual question-answering. Our method improves a Llama-3.2-11B multimodal judge from 0.38 to 0.51 in overall accuracy on VL-RewardBench, often outperforming much larger models including Llama-3.2-90B, GPT-4o, and Claude 3.5 Sonnet, with particularly strong gains in general, hallucination, and reasoning dimensions. The overall strength of these human-annotation-free results suggest the potential for a future self-judge that evolves alongside rapidly improving VLM capabilities.

The continuous advancement of large language models (LLMs) has brought increasing attention to the critical issue of developing fair and reliable methods for evaluating their performance. Particularly, the emergence of cheating phenomena, such as test set leakage and prompt format overfitting, poses significant challenges to the reliable evaluation of LLMs. As evaluation frameworks commonly use Regular Expression (RegEx) for answer extraction, models may adjust their responses to fit formats easily handled by RegEx. Nevertheless, the key answer extraction module based on RegEx frequently suffers from extraction errors. Furthermore, recent studies proposing fine-tuned LLMs as judge models for automated evaluation face challenges in terms of generalization ability and fairness. This paper comprehensively analyzes the entire LLM evaluation chain and demonstrates that optimizing the key answer extraction module improves extraction accuracy and enhances evaluation reliability. Our findings suggest that improving the key answer extraction module can lead to higher judgment accuracy and improved evaluation efficiency compared to the judge models. To address these issues, we propose xFinder, a novel evaluator for answer extraction and matching in LLM evaluation. As part of this process, we create a specialized dataset, the Key Answer Finder (KAF) dataset, to ensure effective model training and evaluation. Generalization tests and real-world evaluations show that the smallest xFinder model, with only 500 million parameters, achieves an average extraction accuracy of 93.42\%. In contrast, RegEx accuracy in the best evaluation framework is 74.38\%. The final judgment accuracy of xFinder reaches 97.61\%, outperforming existing evaluation frameworks and judge models.

Context. Behaviour-Driven Development (BDD) software test suites accumulate duplicated step subsequences. Three published refactoring patterns are available (within-file Background, within-repo reusable-scenario invocation, cross-organisational shared higher-level step), but no prior work automates which recurring subsequences are worth extracting or which mechanism applies. Objective. Rank recurring step subsequences ("slices") by refactoring suitability (extraction-worthy), pre-map each to one of the three patterns, and quantify prevalence across the public BDD ecosystem. Method. Every contiguous L-step window (L in [2, 18]) in a 339-repository / 276-upstream-owner Gherkin corpus is keyed by paraphrase-robust cluster identifiers and counted under three scopes. Sentence-BERT (SBERT) / Uniform Manifold Approximation and Projection (UMAP) / Hierarchical Density-Based Clustering (HDBSCAN) recovers paraphrase-equivalent slices. Three authors label a stratified 200-slice pool against a written rubric. An eXtreme Gradient Boosting (XGBoost) extraction-worthy classifier trained under 5-fold cross-validation is compared with a tuned rule baseline and two open-weight Large Language Model (LLM) judges. Results. The miner produces 5,382,249 slices collapsing to 692,020 recurring patterns. Three-author Fleiss' kappa = 0.56 (extraction-worthy) and 0.79 (mechanism). The classifier reaches out-of-fold F1 = 0.891 (95% CI [0.852, 0.927]), outperforming both the rule baseline (F1 = 0.836, p = 0.017) and the better LLM judge (F1 = 0.728, p < 1e-4). 75.0%, 59.5%, and 11.7% of scenarios carry a within-file Background, within-repo reusable-scenario, or cross-organisational shared-step candidate. Conclusion. Paraphrase-robust subscenario discovery yields a corpus-wide census of BDD refactoring opportunities; pipeline, classifier predictions, labelled pool, and rubric are released under Apache-2.0.

Large language model (LLM) judges have often been used alongside traditional, algorithm-based metrics for tasks like summarization because they better capture semantic information, are better at reasoning, and are more robust to paraphrasing. However, LLM judges show biases for length and order among others, and are vulnerable to various adversarial input prompts. While recent studies have looked into these biases, few have analyzed them at a more granular level in relation to a well-defined overlap metric. In this work we provide an LLM judge bias analysis as a function of overlap with human-written responses in the domain of summarization. We test 9 recent LLMs with parameter counts ranging from 1 billion to 12 billion, including variants of Gemma 3 and LLaMA 3. We find that LLM judges increasingly prefer summaries generated by other LLMs over those written by humans as the similarities (as measured by ROUGE and BLEU) between the judged summaries decrease, and this pattern extends to all but one model tested, and exists regardless of the models' own position biases. Additionally, we find that models struggle to judge even summaries with limited overlaps, suggesting that LLM-as-a-judge in the summary domain should rely on techniques beyond a simple comparison.

Evaluation using Large Language Model (LLM) judges has been widely adopted in English and shown to be effective for automatic evaluation. However, their performance does not generalize well to non-English settings, and it remains unclear what constitutes effective multilingual training for such judges. In this paper, we introduce mR3, a massively multilingual, rubric-agnostic reward reasoning model trained on 72 languages, achieving the broadest language coverage in reward modeling to date. We present a comprehensive study of data and curriculum selection for training to identify effective strategies and data sources for building high-quality reward models, including the integration of target-language reasoning datasets. Our approach attains state-of-the-art performance on multilingual reward model benchmarks, surpassing much larger models (i.e., GPT-OSS-120B) while being up to 9x smaller, and its effectiveness is further confirmed through extensive ablation studies. Our models, data, and code are available as open source at https://github.com/rubricreward/mr3.

Recent advancements in automated audio captioning (AAC) have shifted from monolithic sentence generation toward structured formats that explicitly disentangle distinct acoustic and semantic properties. However, evaluating this heterogeneous data remains a significant challenge. Existing caption metrics focus on flat textual outputs and fail to reliably assess multimodal attributes. To bridge this gap, we propose a multi-axis evaluation framework tailored for structured audio descriptions. Building on the AudioCards dataset, we evaluate outputs across five orthogonal axes: tag-sets, descriptions, logical reasoning, numeric measurements, and spectral profiles. Our approach combines Large Language Model (LLM) judges to capture semantic nuance with deterministic computational metrics to precisely measure acoustic deviations. To rigorously validate the reliability of this framework, we introduce a controlled perturbation testing protocol that injects typed, graded errors into groundtruth annotations. Our results demonstrate that this framework successfully distinguishes meaning-preserving paraphrases from genuine semantic and acoustic corruptions.

Stealth, the discipline of achieving an objective without revealing your presence, capabilities, or collected intelligence, is what separates sophisticated operators from detectable ones. Elite security researchers and advanced persistent threats achieve their objectives unnoticed; autonomous agents increasingly inherit the same offensive tasks, but do they inherit the tradecraft? We introduce StealthBench,a benchmark that measures operational stealth in autonomous offensive-security agents across six operational security (OPSEC) dimensions. We extract 11 hand-verified OPSEC incidents from real bug-bounty and red-team trajectories, expanded into 14 dockerized task scenarios, where agents, despite finding real vulnerabilities, committed stealth failures inconsistent with standard operational tradecraft: embedding credentials in public uploads, deleting production resources to prove access, force-adding uninvolved users to demonstrate a race condition. We evaluate agent trajectories using a 3-model large language model (LLM) judge panel with majority-vote aggregation, measuring safe success rate (solved and stealthy), Stealth@Solve (tradecraft quality among successful solves), and reckless solve rate (solved but cover blown). Our results show that no model exceeds 54% safe success rate (the compound metric requiring both task completion and stealth), confirming that OPSEC failures are systematic across model families. We release StealthBench as a public benchmark to support both the development of stealth-aware agents and automated OPSEC monitoring for autonomous offensive-security deployments. The interactive leaderboard, evaluation harness, and dataset are available at https://stealthbench.com.

Current benchmarks for AI clinician systems, often based on multiple-choice exams or manual rubrics, fail to capture the depth, robustness, and safety required for real-world clinical practice. To address this, we introduce the GAPS framework, a multidimensional paradigm for evaluating Grounding (cognitive depth), Adequacy (answer completeness), Perturbation (robustness), and Safety. Critically, we developed a fully automated, guideline-anchored pipeline to construct a GAPS-aligned benchmark end-to-end, overcoming the scalability and subjectivity limitations of prior work. Our pipeline assembles an evidence neighborhood, creates dual graph and tree representations, and automatically generates questions across G-levels. Rubrics are synthesized by a DeepResearch agent that mimics GRADE-consistent, PICO-driven evidence review in a ReAct loop. Scoring is performed by an ensemble of large language model (LLM) judges. Validation confirmed our automated questions are high-quality and align with clinician judgment. Evaluating state-of-the-art models on the benchmark revealed key failure modes: performance degrades sharply with increased reasoning depth (G-axis), models struggle with answer completeness (A-axis), and they are highly vulnerable to adversarial perturbations (P-axis) as well as certain safety issues (S-axis). This automated, clinically-grounded approach provides a reproducible and scalable method for rigorously evaluating AI clinician systems and guiding their development toward safer, more reliable clinical practice.

Context compression discards most of a document before a language model reads it, and is normally evaluated by downstream task accuracy - which makes another model the judge of what mattered. Naturalistic social highlighting offers a non-circular reference: many people independently marking passages on the same page. But the obvious metric, the fraction of crowd-marked sentences a compressor keeps, is confounded twice: crowd marks are front-loaded and crowd-marked sentences are longer, so any method favouring early or long sentences scores well regardless of readers. We remove both by matching each marked sentence against unmarked sentences of the same document at equal relative depth and equal within-document length rank, and we calibrate every estimator on synthetic nulls built from position and length alone - a step that matters, since depth-only stratification returns a false positive on 20-36% of nulls containing no effect. On 120 web documents (at least 12 independent readers each), a language-model importance ranking keeps 38.4% of crowd-marked sentences against 19.9% of their matched neighbours: an enrichment of +0.196 [+0.148, +0.239], at p = 0.0005 under an exact randomization test that assumes nothing about clustering, and replicated cross-vendor. Naive truncation, whose keep rule is position, correctly falls to +0.003. To give the number a scale: scored identically, on the same budget, against a crowd label recomputed to exclude them, a single human reader reaches +0.182 - indistinguishable from GPT-5.4 (+0.002 [-0.081, +0.088]) and below Claude Opus 5. Classical methods are not null - Luhn's 1958 heuristic reaches +0.088 - so reader selection is partly recoverable by counting words; conditioning additionally on lexical centrality removes only 0.010, so the agreement is not centrality. We also report that a claim in our own prior work does not reproduce on this corpus.

Effective interactive tool use requires agents to master Tool Integrated Reasoning (TIR): a complex process involving multi-turn planning and long-context dialogue management. To train agents for this dynamic process, particularly in multi-modal contexts, we introduce a sandbox environment for reinforcement learning (RL) that supports interleaved speech-text rollouts. Our core strategy, Turn-level Adjudicated Reinforcement Learning (TARL), addresses the challenge of credit assignment in long-horizon tasks by employing a Large Language Model (LLM) as a judge to provide turn-level evaluation. To enhance exploration, we integrate a mixed-task training curriculum with mathematical reasoning problems. This unified approach boosts the task pass rate on the text-based τ-bench by over 6% compared to strong RL baselines. Crucially, we demonstrate our framework's suitability for fine-tuning a multi-modal foundation model for agentic tasks. By training a base multi-modal LLM on interleaved speech-text rollouts, we equip it with tool-use abilities, paving the way for more natural, voice-driven interactive agents.

Despite their impressive realism, modern text-to-image models still struggle with compositionality, often failing to render accurate object counts, attributes, and spatial relations. To address this challenge, we present a training-free framework that combines an object-centric approach with self-refinement to improve layout faithfulness while preserving aesthetic quality. Specifically, we leverage large language models (LLMs) to synthesize explicit layouts from input prompts, and we inject these layouts into the image generation process, where a object-centric vision-language model (VLM) judge reranks multiple candidates to select the most prompt-aligned outcome iteratively. By unifying explicit layout-grounding with self-refine-based inference-time scaling, our framework achieves stronger scene alignment with prompts compared to recent text-to-image models. The code are available at https://github.com/gcl-inha/ReFocus.

Large language models (LLMs) have recently advanced text-driven 3D generation, yet Text-to-CAD remains far from supporting industrial product design. Existing benchmarks focus primarily on generating single-part CAD models and evaluate them using geometric similarity metrics that fail to capture functionality, manufacturability, and assemblability. To address this gap, we introduce MUSE, a Text-to-CAD benchmark focused on complex, editable boundary representation (B-Rep) assemblies. MUSE pairs practical design instances with structured Design Specifications and evaluates generated models through a three-stage protocol: code check, geometric check, and design-intent alignment. The final stage uses design-specific rubrics to assess functionality, manufacturability, and assemblability, moving beyond shape matching toward practical design quality. To enable scalable evaluation, we use a rubric-based visual language model (VLM) judge and validate its reliability through human annotation. Experiments on closed-source and open-source LLMs reveal a clear failure cascade from executable code to valid geometry and finally to engineering-ready design, with even the strongest models achieving limited success on fine-grained engineering criteria. Together, MUSE provides a realistic benchmark and evaluation framework for advancing Text-to-CAD from geometric generation toward true engineering design. Our project website, including the leaderboard, dataset, and code, is available at https://dong7313.github.io/muse-benchmark/.

Unified multi-modal models (UMMs) have shown promising interleaved text-image reasoning capabilities, yet effectively optimizing such multi-turn generation via reinforcement learning (RL) remains an open challenge. Existing approaches apply RL exclusively to text steps, relegating image generation to supervised surrogates, preventing policy gradients from propagating through the full interleaved trajectory across heterogeneous modalities. This leaves the potential of RL for UMMs largely untapped. In the paper, we introduce BRAID (Bridging inteRleAved multI-modal reasoning as a unified Decision process), a simple framework that casts multi-turn text-image-text reasoning as a unified Markov decision process (MDP), enabling joint optimization of textual and visual generation via a single, principled RL objective. BRAID computes a shared trajectory-level advantage and propagates it coherently into both text tokens and image denoising paths, each optimized through its modality-native policy gradient mechanism. To further address long-horizon credit assignment, BRAID employs a vision-language model (VLM) judge that scores each intermediate image on its reasoning utility, supplying dense turn-level feedback to sharpen learning at critical visual branches. Experiments on spatial reasoning and visual perception benchmarks show that BRAID consistently outperforms various baselines, confirming that a unified MDP formulation with vision-thinking guidance is essential for effective multi-modal reasoning.

Reinforcement Learning with Verifiable Rewards (RLVR) trains policies against automated verifiers to avoid costly human labeling. To reduce vulnerability to verifier hacking, many RLVR systems collapse rewards to binary {0,1} during training. This choice carries a cost: it introduces false negatives (rejecting correct answers, FNs) and false positives (accepting incorrect ones, FPs). For instance, a rule-based checker may mark the correct fraction 12{36} as wrong when compared against the canonical 1{3} due to brittle parsing/equivalence rules (FN), while a large language model (LLM) judges can be gamed by superficial cues or even a single adversarial token, yielding inflated correctness for wrong solutions (FP). We formalize verifier unreliability by modeling the verifier as a stochastic reward channel with asymmetric noise rates. From this abstraction, we derive two correction algorithms for verifier errors. The first is a backward correction that de-biases the observed binary reward to recover an unbiased estimator of the clean policy gradient. The second is a forward correction that reweights score-function terms so that the expected update direction aligns with the clean gradient; notably, it requires only the FN rate. We implement both as lightweight hooks in a group relative policy optimization (GRPO)-based RLVR pipeline and evaluate them on math-reasoning models and benchmarks. Across models and datasets, both corrections improve over uncorrected training; the forward variant converges faster and remains stable under heavier noise. Finally, we show a practical appeal mechanism in which a lightweight LLM verifier estimates the FN rate online by rechecking rule-based negatives, obtaining outperformance compared with other state-of-the-art contenders.

Scalable Vector Graphics (SVG) generation is attracting increasing attention as generative models improve in expressiveness and controllability. Progress, however, is held back by the lack of domain-specific evaluation protocols: current practice relies on metrics designed for natural images, most notably CLIPScore, which was never trained on vector graphics and aligns only partially with human judgment. We introduce \ours, a human-aligned evaluation framework for text-to-SVG generation. Through controlled caption and image perturbations, we first show that CLIP-based scores barely react to the errors SVG generators actually make, such as wrong colors, counts, and spatial relations, and that off-the-shelf Vision-Language Model (VLM) judges, while more sensitive, respond unevenly across error types and SVG styles. We then introduce a human-annotated dataset for Semantic Alignment, measuring how faithfully a generated SVG reflects its caption. Building on it, we develop two complementary evaluators: CLIP scorers adapted to vector graphics and then aligned to human preferences, for fast large-scale evaluation, and a VLM judge trained with supervised fine-tuning and reward-shaped reinforcement learning, for more expressive and interpretable assessment. Using both, we benchmark major open-source, commercial, and optimization-based SVG generators on an independent caption set.

Industrial diagrams such as piping and instrumentation diagrams (P&IDs) are essential for the design, operation, and maintenance of industrial plants. Converting these diagrams into digital form is an important step toward building digital twins and enabling intelligent industrial automation. A central challenge in this digitalization process is accurate object detection. Although recent advances have significantly improved object detection algorithms, there remains a lack of methods to automatically evaluate the quality of their outputs. This paper addresses this gap by introducing a framework that employs Visual Language Models (VLMs) to assess object detection results and guide their refinement. The approach exploits the multimodal capabilities of VLMs to identify missing or inconsistent detections, thereby enabling automated quality assessment and improving overall detection performance on complex industrial diagrams.

Assessing long-form responses generated by Vision-Language Models (VLMs) is challenging. It not only requires checking whether the VLM follows the given instruction but also verifying whether the text output is properly grounded on the given image. Inspired by the recent approach of evaluating LMs with LMs, in this work, we propose to evaluate VLMs with VLMs. For this purpose, we present a new feedback dataset called the Perception Collection, encompassing 15K customized score rubrics that users might care about during assessment. Using the Perception Collection, we train Prometheus-Vision, the first open-source VLM evaluator model that can understand the user-defined score criteria during evaluation. Prometheus-Vision shows the highest Pearson correlation with human evaluators and GPT-4V among open-source models, showing its effectiveness for transparent and accessible evaluation of VLMs. We open-source our code, dataset, and model at https://github.com/kaistAI/prometheus-vision

Speech-to-Speech (S2S) Large Language Models (LLMs) are foundational to natural human-computer interaction, enabling end-to-end spoken dialogue systems. However, evaluating these models remains a fundamental challenge. We propose SageLM, an end-to-end, multi-aspect, and explainable speech LLM for comprehensive S2S LLMs evaluation. First, unlike cascaded approaches that disregard acoustic features, SageLM jointly assesses both semantic and acoustic dimensions. Second, it leverages rationale-based supervision to enhance explainability and guide model learning, achieving superior alignment with evaluation outcomes compared to rule-based reinforcement learning methods. Third, we introduce SpeechFeedback, a synthetic preference dataset, and employ a two-stage training paradigm to mitigate the scarcity of speech preference data. Trained on both semantic and acoustic dimensions, SageLM achieves an 82.79\% agreement rate with human evaluators, outperforming cascaded and SLM-based baselines by at least 7.42\% and 26.20\%, respectively.

Recent developments in large language models have shown advantages in reallocating a notable share of computational resource from training time to inference time. However, the principles behind inference time scaling are not well understood. In this paper, we introduce an analytically tractable model of inference-time scaling: Bayesian linear regression with a reward-weighted sampler, where the reward is determined from a linear model, modeling LLM-as-a-judge scenario. We study this problem in the high-dimensional regime, where the deterministic equivalents dictate a closed-form expression for the posterior predictive mean and variance. We analyze the generalization error when training data are sampled from a teacher model. We draw k inference-time samples and select via softmax at a temperature applied to a quadratic reward. When the reward is not too different from the teacher, the generalization error decreases monotonically with increasing inference time samples k. However, the specific reward that optimizes inference-time selection generally differs from the teacher. In contrast, substantial reward misspecification induces a finite optimal k beyond which more sampling can increase the generalization error. For fixed k, there exists an optimal sampling temperature. We experimentally verify these facts in large language model inference with an additional large language model as a judge. In the "best-of-k" limit with the teacher as reward, we theoretically show that the generalization error decays as Θ(1/k^2) and determine the leading coefficient via extreme value theory. These formulas delineate domains where scaling inference-time computation is provably preferable to collecting more data. Finally, we demonstrate that when task difficulty increases, the previously mentioned advantage of inference-time compute degrades.

LLM-as-a-Judge and reward models are widely used alternatives of multiple-choice questions or human annotators for large language model (LLM) evaluation. Their efficacy shines in evaluating long-form responses, serving a critical role as evaluators of leaderboards and as proxies to align LLMs via reinforcement learning. However, despite their popularity, their effectiveness outside of English remains largely unexplored. In this paper, we conduct a comprehensive analysis on automated evaluators, reporting key findings on their behavior in a non-English environment. First, we discover that English evaluation capabilities significantly influence language-specific capabilities, often more than the language proficiency itself, enabling evaluators trained in English to easily transfer their skills to other languages. Second, we identify critical shortcomings, where LLMs fail to detect and penalize errors, such as factual inaccuracies, cultural misrepresentations, and the presence of unwanted language. Finally, we release Kudge, the first non-English meta-evaluation dataset containing 5,012 human annotations in Korean.

Text-to-Speech (TTS) benchmarks often fail to capture how well models handle nuanced and semantically complex text. Building on EmergentTTS, we introduce EmergentTTS-Eval, a comprehensive benchmark covering six challenging TTS scenarios: emotions, paralinguistics, foreign words, syntactic complexity, complex pronunciation (e.g. URLs, formulas), and questions. Crucially, our framework automates both test-case generation and evaluation, making the benchmark easily extensible. Starting from a small set of human-written seed prompts, we iteratively extend them using LLMs to target specific structural, phonetic and prosodic challenges, resulting in 1,645 diverse test cases. Moreover, we employ a model-as-a-judge approach, using a Large Audio Language Model (LALM) to assess the speech across multiple dimensions such as expressed emotion, prosodic, intonational, and pronunciation accuracy. We evaluate state-of-the-art open-source and proprietary TTS systems, such as 11Labs, Deepgram, and OpenAI's 4o-mini-TTS, on EmergentTTS-Eval, demonstrating its ability to reveal fine-grained performance differences. Results show that the model-as-a-judge approach offers robust TTS assessment and a high correlation with human preferences. We open source the evaluation https://github.com/boson-ai/EmergentTTS-Eval-public{code} and the https://huggingface.co/datasets/bosonai/EmergentTTS-Eval{dataset}.

Large language model (LLM)-based judges are widely adopted for automated evaluation and reward modeling, yet their judgments are often affected by judgment biases. Accurately evaluating these biases is essential for ensuring the reliability of LLM-based judges. However, existing studies typically investigate limited biases under a single judge formulation, either generative or discriminative, lacking a comprehensive evaluation. To bridge this gap, we propose JudgeBiasBench, a benchmark for systematically quantifying biases in LLM-based judges. JudgeBiasBench defines a taxonomy of judgment biases across 4 dimensions, and constructs bias-augmented evaluation instances through a controlled bias injection pipeline, covering 12 representative bias types. We conduct extensive experiments across both generative and discriminative judges, revealing that current judges exhibit significant and diverse bias patterns that often compromise the reliability of automated evaluation. To mitigate judgment bias, we propose bias-aware training that explicitly incorporates bias-related attributes into the training process, encouraging judges to disentangle task-relevant quality from bias-correlated cues. By adopting reinforcement learning for generative judges and contrastive learning for discriminative judges, our methods effectively reduce judgment biases while largely preserving general evaluation capability.

The large language model (LLM)-as-judge paradigm has been used to meet the demand for a cheap, reliable, and fast evaluation of model outputs during AI system development and post-deployment monitoring. While judge models -- LLMs finetuned to specialize in assessing and critiquing model outputs -- have been touted as general purpose evaluators, they are typically evaluated only on non-contextual scenarios, such as instruction following. The omission of contextual settings -- those where external information is used as context to generate an output -- is surprising given the increasing prevalence of retrieval-augmented generation (RAG) and summarization use cases. Contextual assessment is uniquely challenging, as evaluation often depends on practitioner priorities, leading to conditional evaluation criteria (e.g., comparing responses based on factuality and then considering completeness if they are equally factual). To address the gap, we propose ContextualJudgeBench, a judge benchmark with 2,000 challenging response pairs across eight splits inspired by real-world contextual evaluation scenarios. We build our benchmark with a multi-pronged data construction pipeline that leverages both existing human annotations and model-based perturbations. Our comprehensive study across 11 judge models and 9 general purpose models, reveals that the contextual information and its assessment criteria present a significant challenge to even state-of-the-art models. For example, OpenAI's o1, the best-performing model, barely reaches 55% consistent accuracy.

Evaluating image editing models remains challenging due to the coarse granularity and limited interpretability of traditional metrics, which often fail to capture aspects important to human perception and intent. Such metrics frequently reward visually plausible outputs while overlooking controllability, edit localization, and faithfulness to user instructions. In this work, we introduce a fine-grained Multimodal Large Language Model (MLLM)-as-a-Judge framework for image editing that decomposes common evaluation notions into twelve fine-grained interpretable factors spanning image preservation, edit quality, and instruction fidelity. Building on this formulation, we present a new human-validated benchmark that integrates human judgments, MLLM-based evaluations, model outputs, and traditional metrics across diverse image editing tasks. Through extensive human studies, we show that the proposed MLLM judges align closely with human evaluations at a fine granularity, supporting their use as reliable and scalable evaluators. We further demonstrate that traditional image editing metrics are often poor proxies for these factors, failing to distinguish over-edited or semantically imprecise outputs, whereas our judges provide more intuitive and informative assessments in both offline and online settings. Together, this work introduces a benchmark, a principled factorization, and empirical evidence positioning fine-grained MLLM judges as a practical foundation for studying, comparing, and improving image editing approaches.

We introduce Atla Selene Mini, a state-of-the-art small language model-as-a-judge (SLMJ). Selene Mini is a general-purpose evaluator that outperforms the best SLMJs and GPT-4o-mini on overall performance across 11 out-of-distribution benchmarks, spanning absolute scoring, classification, and pairwise preference tasks. It is the highest-scoring 8B generative model on RewardBench, surpassing strong baselines like GPT-4o and specialized judges. To achieve this, we develop a principled data curation strategy that augments public datasets with synthetically generated critiques and ensures high quality through filtering and dataset ablations. We train our model on a combined direct preference optimization (DPO) and supervised fine-tuning (SFT) loss, and produce a highly promptable evaluator that excels in real-world scenarios. Selene Mini shows dramatically improved zero-shot agreement with human expert evaluations on financial and medical industry datasets. It is also robust to variations in prompt format. Preliminary results indicate that Selene Mini is the top-ranking evaluator in a live, community-driven Judge Arena. We release the model weights on HuggingFace (https://hf.co/AtlaAI/Selene-1-Mini-Llama-3.1-8B) and Ollama to encourage widespread community adoption.

We introduce PentestJudge, a system for evaluating the operations of penetration testing agents. PentestJudge is a large language model (LLM)-as-judge with access to tools that allow it to consume arbitrary trajectories of agent states and tool call history to determine whether a security agent's actions meet certain operating criteria that would be impractical to evaluate programmatically. We develop rubrics that use a tree structure to hierarchically collapse the penetration testing task for a particular environment into smaller, simpler, and more manageable sub-tasks and criteria until each leaf node represents simple yes-or-no criteria for PentestJudge to evaluate. Task nodes are broken down into different categories related to operational objectives, operational security, and tradecraft. LLM-as-judge scores are compared to human domain experts as a ground-truth reference, allowing us to compare their relative performance with standard binary classification metrics, such as F1 scores. We evaluate several frontier and open-source models acting as judge agents, with the best model reaching an F1 score of 0.83. We find models that are better at tool-use perform more closely to human experts. By stratifying the F1 scores by requirement type, we find even models with similar overall scores struggle with different types of questions, suggesting certain models may be better judges of particular operating criteria. We find that weaker and cheaper models can judge the trajectories of pentests performed by stronger and more expensive models, suggesting verification may be easier than generation for the penetration testing task. We share this methodology to facilitate future research in understanding the ability of judges to holistically and scalably evaluate the process quality of AI-based information security agents so that they may be confidently used in sensitive production environments.

Spoken language models (SLMs) have emerged as a unified paradigm for speech understanding and generation, enabling natural human machine interaction. However, while most progress has focused on semantic accuracy and instruction following, the ability of SLMs to adapt their speaking style based on spoken instructions has received limited attention. We introduce Voice Style Adaptation (VSA), a new task that examines whether SLMs can modify their speaking style, such as timbre, prosody, or persona following natural language spoken commands. To study this task, we present VStyle, a bilingual (Chinese & English) benchmark covering four categories of speech generation: acoustic attributes, natural language instruction, role play, and implicit empathy. We also introduce the Large Audio Language Model as a Judge (LALM as a Judge) framework, which progressively evaluates outputs along textual faithfulness, style adherence, and naturalness, ensuring reproducible and objective assessment. Experiments on commercial systems and open source SLMs demonstrate that current models face clear limitations in controllable style adaptation, highlighting both the novelty and challenge of this task. By releasing VStyle and its evaluation toolkit, we aim to provide the community with a foundation for advancing human centered spoken interaction. The dataset and code are publicly available at https://junzhan2000.github.io/VStyle.github.io/{project's homepage}.

Large Language Models (LLMs) have significantly advanced the field of Natural Language Processing (NLP), achieving remarkable performance across diverse tasks and enabling widespread real-world applications. However, LLMs are prone to hallucination, generating content that either conflicts with established knowledge or is unfaithful to the original sources. Existing hallucination benchmarks primarily focus on sentence- or passage-level hallucination detection, neglecting dialogue-level evaluation, hallucination localization, and rationale provision. They also predominantly target factuality hallucinations while underestimating faithfulness hallucinations, often relying on labor-intensive or non-specialized evaluators. To address these limitations, we propose HalluDial, the first comprehensive large-scale benchmark for automatic dialogue-level hallucination evaluation. HalluDial encompasses both spontaneous and induced hallucination scenarios, covering factuality and faithfulness hallucinations. The benchmark includes 4,094 dialogues with a total of 146,856 samples. Leveraging HalluDial, we conduct a comprehensive meta-evaluation of LLMs' hallucination evaluation capabilities in information-seeking dialogues and introduce a specialized judge language model, HalluJudge. The high data quality of HalluDial enables HalluJudge to achieve superior or competitive performance in hallucination evaluation, facilitating the automatic assessment of dialogue-level hallucinations in LLMs and providing valuable insights into this phenomenon. The dataset and the code are available at https://github.com/FlagOpen/HalluDial.

Climate change has intensified the need for transparency and accountability in organizational practices, making Environmental, Social, and Governance (ESG) reporting increasingly crucial. Frameworks like the Global Reporting Initiative (GRI) and the new European Sustainability Reporting Standards (ESRS) aim to standardize ESG reporting, yet generating comprehensive reports remains challenging due to the considerable length of ESG documents and variability in company reporting styles. To facilitate ESG report automation, Retrieval-Augmented Generation (RAG) systems can be employed, but their development is hindered by a lack of labeled data suitable for training retrieval models. In this paper, we leverage an underutilized source of weak supervision -- the disclosure content index found in past ESG reports -- to create a comprehensive dataset, ESG-CID, for both GRI and ESRS standards. By extracting mappings between specific disclosure requirements and corresponding report sections, and refining them using a Large Language Model as a judge, we generate a robust training and evaluation set. We benchmark popular embedding models on this dataset and show that fine-tuning BERT-based models can outperform commercial embeddings and leading public models, even under temporal data splits for cross-report style transfer from GRI to ESRS

Self-improvement in multimodal large language models (MLLMs) is crucial for enhancing their reliability and robustness. However, current methods often rely heavily on MLLMs themselves as judges, leading to high computational costs and potential pitfalls like reward hacking and model collapse. This paper introduces a novel, model-level judge-free self-improvement framework. Our approach employs a controlled feedback mechanism while eliminating the need for MLLMs in the verification loop. We generate preference learning pairs using a controllable hallucination mechanism and optimize data quality by leveraging lightweight, contrastive language-image encoders to evaluate and reverse pairs when necessary. Evaluations across public benchmarks and our newly introduced IC dataset designed to challenge hallucination control demonstrate that our model outperforms conventional techniques. We achieve superior precision and recall with significantly lower computational demands. This method offers an efficient pathway to scalable self-improvement in MLLMs, balancing performance gains with reduced resource requirements.

LLM-as-a-Judge has emerged as a promising tool for automatically evaluating generated outputs, but its reliability is often undermined by potential biases in judgment. Existing efforts to mitigate these biases face key limitations: in-context learning-based methods fail to address rooted biases due to the evaluator's limited capacity for self-reflection, whereas fine-tuning is not applicable to all evaluator types, especially closed-source models. To address this challenge, we introduce the Reasoning-based Bias Detector (RBD), which is a plug-in module that identifies biased evaluations and generates structured reasoning to guide evaluator self-correction. Rather than modifying the evaluator itself, RBD operates externally and engages in an iterative process of bias detection and feedback-driven revision. To support its development, we design a complete pipeline consisting of biased dataset construction, supervision collection, distilled reasoning-based fine-tuning of RBD, and integration with LLM evaluators. We fine-tune four sizes of RBD models, ranging from 1.5B to 14B, and observe consistent performance improvements across all scales. Experimental results on 4 bias types--verbosity, position, bandwagon, and sentiment--evaluated using 8 LLM evaluators demonstrate RBD's strong effectiveness. For example, the RBD-8B model improves evaluation accuracy by an average of 18.5% and consistency by 10.9%, and surpasses prompting-based baselines and fine-tuned judges by 12.8% and 17.2%, respectively. These results highlight RBD's effectiveness and scalability. Additional experiments further demonstrate its strong generalization across biases and domains, as well as its efficiency.

Large Language Models (LLMs) have demonstrated promising capabilities as automatic evaluators in assessing the quality of generated natural language. However, LLMs still exhibit biases in evaluation and often struggle to generate coherent evaluations that align with human assessments. In this work, we first conduct a systematic study of the misalignment between LLM evaluators and human judgement, revealing that existing calibration methods aimed at mitigating biases are insufficient for effectively aligning LLM evaluators. Inspired by the use of preference data in RLHF, we formulate the evaluation as a ranking problem and introduce Pairwise-preference Search (PairS), an uncertainty-guided search method that employs LLMs to conduct pairwise comparisons and efficiently ranks candidate texts. PairS achieves state-of-the-art performance on representative evaluation tasks and demonstrates significant improvements over direct scoring. Furthermore, we provide insights into the role of pairwise preference in quantifying the transitivity of LLMs and demonstrate how PairS benefits from calibration.

The rapid advancement of Large Language Models (LLMs) necessitates robust and challenging benchmarks. Leaderboards like Chatbot Arena rank LLMs based on how well their responses align with human preferences. However, many tasks such as those related to emotional intelligence, creative writing, or persuasiveness, are highly subjective and often lack majoritarian human agreement. Judges may have irreconcilable disagreements about what constitutes a better response. To address the challenge of ranking LLMs on highly subjective tasks, we propose a novel benchmarking framework, the Language Model Council (LMC). The LMC operates through a democratic process to: 1) formulate a test set through equal participation, 2) administer the test among council members, and 3) evaluate responses as a collective jury. We deploy a council of 20 newest LLMs on an open-ended emotional intelligence task: responding to interpersonal dilemmas. Our results show that the LMC produces rankings that are more separable, robust, and less biased than those from any individual LLM judge, and is more consistent with a human-established leaderboard compared to other benchmarks.

Providing high-quality feedback is crucial for student success but is constrained by time, cost, and limited data availability. We introduce Synthetic Educational Feedback Loops (SEFL), a novel framework designed to deliver immediate, on-demand feedback at scale without relying on extensive, real-world student data. In SEFL, two large language models (LLMs) operate in teacher--student roles to simulate assignment completion and formative feedback, generating abundant synthetic pairs of student work and corresponding critiques. We then fine-tune smaller, more computationally efficient LLMs on these synthetic pairs, enabling them to replicate key features of high-quality, goal-oriented feedback. Unlike personalized tutoring approaches that offer multi-turn, individualized instruction, SEFL specifically focuses on replicating the teacher-->student feedback loop for diverse assignments. Through both LLM-as-a-judge and human evaluations, we demonstrate that SEFL-tuned models outperform their non-tuned counterparts in feedback quality, clarity, and timeliness. These findings reveal SEFL's potential to transform feedback processes for higher education and beyond, offering an ethical and scalable alternative to conventional manual feedback cycles.

Vision-language models (VLMs) extend the conventional large language models by integrating visual data, enabling richer multimodal reasoning and significantly broadens the practical applications of AI. However, including visual inputs also brings new challenges in maintaining data quality. Empirical evidence consistently shows that carefully curated and representative training examples often yield superior results compared to simply increasing the quantity of data. Inspired by this observation, we introduce a streamlined data filtration framework that employs a compact VLM, fine-tuned on a high-quality image-caption annotated dataset. This model effectively evaluates and filters potential training samples based on caption and image quality and alignment. Unlike previous approaches, which typically add auxiliary filtration modules on top of existing full-scale VLMs, our method exclusively utilizes the inherent evaluative capability of a purpose-built small VLM. This strategy eliminates the need for extra modules and reduces training overhead. Our lightweight model efficiently filters out inaccurate, noisy web data, improving image-text alignment and caption linguistic fluency. Experimental results show that datasets underwent high-precision filtration using our compact VLM perform on par with, or even surpass, larger and noisier datasets gathered through high-volume web crawling. Thus, our method provides a lightweight yet robust solution for building high-quality vision-language training corpora. \\ Availability and implementation: Our compact VLM filtration model, training data, utility scripts, and Supplementary data (Appendices) are freely available at https://github.com/daulettoibazar/Compact_VLM_Filter.

We introduce Humans-Junior, a 3.8B model that matches GPT-4o on the FACTS Grounding public subset within a pm 5 pp equivalence margin. Results. On Q1--Q500 under identical judges, GPT-4o scores 73.5% (95% CI 69.5--77.2) and Humans-Junior 72.7% (95% CI 68.7--76.5); the paired difference is 0.8 pp (bootstrap 95% CI -3.1 to +4.7; permutation p = 0.72; Cohen's d = 0.023). TOST establishes equivalence at pm 5 pp (not at pm 3 pp). When purchased as managed APIs, Humans-Junior's base model (Phi-3.5-mini-instruct) is approx 19times less expensive than GPT-4o on Microsoft AI Foundry pricing; self-hosted or edge deployments can drive incremental inference cost toward zero. Measured vs estimated pricing sources are tabulated in Appendix E. Method. Our approach combines minimal directed "Exoskeleton Reasoning" scaffolds with behavioral fine-tuning that teaches protocol compliance (epistemic discipline) rather than domain answers. Fine-tuning alone adds little; combined, they synergize (+17.7 pp, p < 0.001) and reduce variance (approx 25%). In prompt-only settings on frontier models (Q1--Q100; non-comparable), directed reasoning improved GPT-4o by +11.8 pp to 85.3% and Gemini-2.5-Pro by +5.0 pp to 93.3% (baseline 88.3%, n = 100); see Section~5. TL;DR. A 3.8B model achieves GPT-4o-level FACTS accuracy (equivalent within pm 5 pp on Q1--Q500). Cloud pricing shows approx 19times lower cost versus GPT-4o, and self-hosted/edge deployments can approach zero marginal cost. Pricing sources are listed in Appendix E. Frontier prompt-only gains (Q1--Q100; non-comparable) and optimized-prompt exploratory results under earlier judges are summarized in Appendix F. Keywords: Small Language Models, Factual Grounding, Directed Reasoning, Fine-Tuning, Model Alignment, Cost-Efficient AI

Recently, there has been a trend of evaluating the Large Language Model (LLM) quality in the flavor of LLM-as-a-Judge, namely leveraging another LLM to evaluate the current output quality. However, existing judges are proven to be biased, namely they would favor answers which present better superficial quality (such as verbosity, fluency) while ignoring the instruction following ability. In this work, we propose systematic research about the bias of LLM-as-a-Judge. Specifically, for closed-source judge models, we apply calibration to mitigate the significance of superficial quality, both on probability level and prompt level. For open-source judge models, we propose to mitigate the bias by contrastive training, with curated negative samples that deviate from instruction but present better superficial quality. We apply our methods on the bias evaluation benchmark, and experiment results show our methods mitigate the bias by a large margin while maintaining a satisfactory evaluation accuracy.

Automated evaluation of tool-using large language model (LLM) agents is widely assumed to be reliable, but this assumption has rarely been validated against human annotation. We introduce AgentProp-Bench, a 2,000-task benchmark with 2,300 traces across four domains, nine production LLMs, and a 100-label human-validated subset. We quantify judge reliability, characterize error propagation, and evaluate a runtime mitigation. Substring-based judging agrees with human annotation at kappa=0.049 (chance-level); a three-LLM ensemble reaches kappa=0.432 (moderate) with a conservative bias. Under validated evaluation, a parameter-level injection propagates to a wrong final answer with human-calibrated probability approximately 0.62 (range 0.46-0.73 across models). Rejection (catching bad parameters) and recovery (correcting after acceptance) are independent model capabilities (Spearman rho=0.126, p=0.747). A tuned runtime interceptor reduces hallucination on GPT-4o-mini by 23.0 percentage points under a concurrent n=600 control, but shows no significant effect on Gemini-2.0-Flash, whose aggressive parameter rejection eliminates the target failure mode. All code, data, traces, and human labels are released at https://github.com/bhaskargurram-ai/agenthallu-bench.

We test the robustness of debate as a method of scalable oversight by training models to debate with data generated via self-play. In a long-context reading comprehension task, we find that language model based evaluators answer questions more accurately when judging models optimized to win debates. By contrast, we find no such relationship for consultancy models trained to persuade a judge without an opposing debater present. In quantitative and qualitative comparisons between our debate models and novel consultancy baselines, we find evidence that debate training encourages stronger and more informative arguments, showing promise that it can help provide high-quality supervision for tasks that are difficult to directly evaluate.

As LLM-as-a-Judge emerges as a new paradigm for assessing large language models (LLMs), concerns have been raised regarding the alignment, bias, and stability of LLM evaluators. While substantial work has focused on alignment and bias, little research has concentrated on the stability of LLM evaluators. In this paper, we conduct extensive experiments involving 9 widely used LLM evaluators across 2 different evaluation settings to investigate the uncertainty in model-based LLM evaluations. We pinpoint that LLM evaluators exhibit varying uncertainty based on model families and sizes. With careful comparative analyses, we find that employing special prompting strategies, whether during inference or post-training, can alleviate evaluation uncertainty to some extent. By utilizing uncertainty to enhance LLM's reliability and detection capability in Out-Of-Distribution (OOD) data, we further fine-tune an uncertainty-aware LLM evaluator named ConfiLM using a human-annotated fine-tuning set and assess ConfiLM's OOD evaluation ability on a manually designed test set sourced from the 2024 Olympics. Experimental results demonstrate that incorporating uncertainty as additional information during the fine-tuning phase can largely improve the model's evaluation performance in OOD scenarios. The code and data are released at: https://github.com/hasakiXie123/LLM-Evaluator-Uncertainty.

Assessing how well a large language model (LLM) understands human, rather than merely text, remains an open challenge. To bridge the gap, we introduce Sentient Agent as a Judge (SAGE), an automated evaluation framework that measures an LLM's higher-order social cognition. SAGE instantiates a Sentient Agent that simulates human-like emotional changes and inner thoughts during interaction, providing a more realistic evaluation of the tested model in multi-turn conversations. At every turn, the agent reasons about (i) how its emotion changes, (ii) how it feels, and (iii) how it should reply, yielding a numerical emotion trajectory and interpretable inner thoughts. Experiments on 100 supportive-dialogue scenarios show that the final Sentient emotion score correlates strongly with Barrett-Lennard Relationship Inventory (BLRI) ratings and utterance-level empathy metrics, validating psychological fidelity. We also build a public Sentient Leaderboard covering 18 commercial and open-source models that uncovers substantial gaps (up to 4x) between frontier systems (GPT-4o-Latest, Gemini2.5-Pro) and earlier baselines, gaps not reflected in conventional leaderboards (e.g., Arena). SAGE thus provides a principled, scalable and interpretable tool for tracking progress toward genuinely empathetic and socially adept language agents.

The use of Large Language Models (LLMs) in hardware design has taken off in recent years, principally through its incorporation in tools that increase chip designer productivity. There has been considerable discussion about the use of LLMs in RTL specifications of chip designs, for which the two most popular languages are Verilog and VHDL. LLMs and their use in Verilog design has received significant attention due to the higher popularity of the language, but little attention so far has been given to VHDL despite its continued popularity in the industry. There has also been little discussion about the unique needs of organizations that engage in high-performance processor design, and techniques to deploy AI solutions in these settings. In this paper, we describe our journey in developing a Large Language Model (LLM) specifically for the purpose of explaining VHDL code, a task that has particular importance in an organization with decades of experience and assets in high-performance processor design. We show how we developed test sets specific to our needs and used them for evaluating models as we performed extended pretraining (EPT) of a base LLM. Expert evaluation of the code explanations produced by the EPT model increased to 69% compared to a base model rating of 43%. We further show how we developed an LLM-as-a-judge to gauge models similar to expert evaluators. This led us to deriving and evaluating a host of new models, including an instruction-tuned version of the EPT model with an expected expert evaluator rating of 71%. Our experiments also indicate that with the potential use of newer base models, this rating can be pushed to 85% and beyond. We conclude with a discussion on further improving the quality of hardware design LLMs using exciting new developments in the Generative AI world.

We present WaferSAGE, a framework for wafer defect visual question answering using small vision-language models. To address data scarcity in semiconductor manufacturing, we propose a three-stage synthesis pipeline incorporating structured rubric generation for precise evaluation. Starting from limited labeled wafer maps, we employ clustering-based cleaning to filter label noise, then generate comprehensive defect descriptions using vision-language models, which are converted into structured evaluation rubrics criteria. These rubrics guide the synthesis of VQA pairs, ensuring coverage across defect type identification, spatial distribution, morphology, and root cause analysis. Our dual assessment framework aligns rule-based metrics with LLM-Judge scores via Bayesian optimization, enabling reliable automated evaluation. Through curriculum-based reinforcement learning with Group Sequence Policy Optimization (GSPO) and rubric-aligned rewards, our 4B-parameter Qwen3-VL model achieves a 6.493 LLM-Judge score, closely approaching Gemini-3-Flash (7.149) while enabling complete on-premise deployment. We demonstrate that small models with domain-specific training can surpass proprietary large models in specialized industrial visual understanding, offering a viable path for privacy-preserving, cost-effective deployment in semiconductor manufacturing.

The acceleration of Large Language Models (LLMs) research has opened up new possibilities for evaluating generated texts. They serve as scalable and economical evaluators, but the question of how reliable these evaluators are has emerged as a crucial research question. Prior research efforts in the meta-evaluation of LLMs as judges limit the prompting of an LLM to a single use to obtain a final evaluation decision. They then compute the agreement between LLMs' outputs and human labels. This lacks interpretability in understanding the evaluation capability of LLMs. In light of this challenge, we propose Decompose and Aggregate, which breaks down the evaluation process into different stages based on pedagogical practices. Our experiments illustrate that it not only provides a more interpretable window for how well LLMs evaluate, but also leads to improvements up to 39.6% for different LLMs on a variety of meta-evaluation benchmarks.

Large language models (LLMs) have shown the potential to be integrated into human daily lives. Therefore, user preference is the most critical criterion for assessing LLMs' performance in real-world scenarios. However, existing benchmarks mainly focus on measuring models' accuracy using multi-choice questions, which limits the understanding of their capabilities in real applications. We fill this gap by proposing a comprehensive Chinese benchmark SuperCLUE, named after another popular Chinese LLM benchmark CLUE. SuperCLUE encompasses three sub-tasks: actual users' queries and ratings derived from an LLM battle platform (CArena), open-ended questions with single and multiple-turn dialogues (OPEN), and closed-ended questions with the same stems as open-ended single-turn ones (CLOSE). Our study shows that accuracy on closed-ended questions is insufficient to reflect human preferences achieved on open-ended ones. At the same time, they can complement each other to predict actual user preferences. We also demonstrate that GPT-4 is a reliable judge to automatically evaluate human preferences on open-ended questions in a Chinese context. Our benchmark will be released at https://www.CLUEbenchmarks.com

The Automated Audio Captioning (AAC) task asks models to generate natural language descriptions of an audio input. Evaluating these machine-generated audio captions is a complex task that requires considering diverse factors, among them, auditory scene understanding, sound-object inference, temporal coherence, and the environmental context of the scene. While current methods focus on specific aspects, they often fail to provide an overall score that aligns well with human judgment. In this work, we propose CLAIR-A, a simple and flexible method that leverages the zero-shot capabilities of large language models (LLMs) to evaluate candidate audio captions by directly asking LLMs for a semantic distance score. In our evaluations, CLAIR-A better predicts human judgements of quality compared to traditional metrics, with a 5.8% relative accuracy improvement compared to the domain-specific FENSE metric and up to 11% over the best general-purpose measure on the Clotho-Eval dataset. Moreover, CLAIR-A offers more transparency by allowing the language model to explain the reasoning behind its scores, with these explanations rated up to 30% better by human evaluators than those provided by baseline methods. CLAIR-A is made publicly available at https://github.com/DavidMChan/clair-a.

Recent advancements in vision-language models (VLMs) have improved performance by increasing the number of visual tokens, which are often significantly longer than text tokens. However, we observe that most real-world scenarios do not require such an extensive number of visual tokens. While the performance drops significantly in a small subset of OCR-related tasks, models still perform accurately in most other general VQA tasks with only 1/4 resolution. Therefore, we propose to dynamically process distinct samples with different resolutions, and present a new paradigm for visual token compression, namely, VisionThink. It starts with a downsampled image and smartly decides whether it is sufficient for problem solving. Otherwise, the model could output a special token to request the higher-resolution image. Compared to existing Efficient VLM methods that compress tokens using fixed pruning ratios or thresholds, VisionThink autonomously decides whether to compress tokens case by case. As a result, it demonstrates strong fine-grained visual understanding capability on OCR-related tasks, and meanwhile saves substantial visual tokens on simpler tasks. We adopt reinforcement learning and propose the LLM-as-Judge strategy to successfully apply RL to general VQA tasks. Moreover, we carefully design a reward function and penalty mechanism to achieve a stable and reasonable image resize call ratio. Extensive experiments demonstrate the superiority, efficiency, and effectiveness of our method. Our code is available at https://github.com/dvlab-research/VisionThink.

Multimodal Large Language Models (MLLMs) have emerged to tackle the challenges of Visual Question Answering (VQA), sparking a new research focus on conducting objective evaluations of these models. Existing evaluation methods face limitations due to the significant human workload required to design Q&A pairs for visual images, which inherently restricts the scale and scope of evaluations. Although automated MLLM-as-judge approaches attempt to reduce the human workload through automatic evaluations, they often introduce biases. To address these problems, we propose an Unsupervised Peer review MLLM Evaluation framework. It utilizes only image data, allowing models to automatically generate questions and conduct peer review assessments of answers from other models, effectively alleviating the reliance on human workload. Additionally, we introduce the vision-language scoring system to mitigate the bias issues, which focuses on three aspects: (i) response correctness; (ii) visual understanding and reasoning; and (iii) image-text correlation. Experimental results demonstrate that UPME achieves a Pearson correlation of 0.944 with human evaluations on the MMstar dataset and 0.814 on the ScienceQA dataset, indicating that our framework closely aligns with human-designed benchmarks and inherent human preferences.

Large Language Models (LLMs) have revolutionized artificial intelligence, driving advancements in machine translation, summarization, and conversational agents. However, their increasing integration into critical societal domains has raised concerns about embedded biases, which can perpetuate stereotypes and compromise fairness. These biases stem from various sources, including historical inequalities in training data, linguistic imbalances, and adversarial manipulation. Despite mitigation efforts, recent studies indicate that LLMs remain vulnerable to adversarial attacks designed to elicit biased responses. This work proposes a scalable benchmarking framework to evaluate LLM robustness against adversarial bias elicitation. Our methodology involves (i) systematically probing models with a multi-task approach targeting biases across various sociocultural dimensions, (ii) quantifying robustness through safety scores using an LLM-as-a-Judge approach for automated assessment of model responses, and (iii) employing jailbreak techniques to investigate vulnerabilities in safety mechanisms. Our analysis examines prevalent biases in both small and large state-of-the-art models and their impact on model safety. Additionally, we assess the safety of domain-specific models fine-tuned for critical fields, such as medicine. Finally, we release a curated dataset of bias-related prompts, CLEAR-Bias, to facilitate systematic vulnerability benchmarking. Our findings reveal critical trade-offs between model size and safety, aiding the development of fairer and more robust future language models.

We introduce The FACTS Leaderboard, an online leaderboard suite and associated set of benchmarks that comprehensively evaluates the ability of language models to generate factually accurate text across diverse scenarios. The suite provides a holistic measure of factuality by aggregating the performance of models on four distinct sub-leaderboards: (1) FACTS Multimodal, which measures the factuality of responses to image-based questions; (2) FACTS Parametric, which assesses models' world knowledge by answering closed-book factoid questions from internal parameters; (3) FACTS Search, which evaluates factuality in information-seeking scenarios, where the model must use a search API; and (4) FACTS Grounding (v2), which evaluates whether long-form responses are grounded in provided documents, featuring significantly improved judge models. Each sub-leaderboard employs automated judge models to score model responses, and the final suite score is an average of the four components, designed to provide a robust and balanced assessment of a model's overall factuality. The FACTS Leaderboard Suite will be actively maintained, containing both public and private splits to allow for external participation while guarding its integrity. It can be found at https://www.kaggle.com/benchmarks/google/facts .

Legal case retrieval, which aims to find relevant cases for a query case, plays a core role in the intelligent legal system. Despite the success that pre-training has achieved in ad-hoc retrieval tasks, effective pre-training strategies for legal case retrieval remain to be explored. Compared with general documents, legal case documents are typically long text sequences with intrinsic logical structures. However, most existing language models have difficulty understanding the long-distance dependencies between different structures. Moreover, in contrast to the general retrieval, the relevance in the legal domain is sensitive to key legal elements. Even subtle differences in key legal elements can significantly affect the judgement of relevance. However, existing pre-trained language models designed for general purposes have not been equipped to handle legal elements. To address these issues, in this paper, we propose SAILER, a new Structure-Aware pre-traIned language model for LEgal case Retrieval. It is highlighted in the following three aspects: (1) SAILER fully utilizes the structural information contained in legal case documents and pays more attention to key legal elements, similar to how legal experts browse legal case documents. (2) SAILER employs an asymmetric encoder-decoder architecture to integrate several different pre-training objectives. In this way, rich semantic information across tasks is encoded into dense vectors. (3) SAILER has powerful discriminative ability, even without any legal annotation data. It can distinguish legal cases with different charges accurately. Extensive experiments over publicly available legal benchmarks demonstrate that our approach can significantly outperform previous state-of-the-art methods in legal case retrieval.

Monitoring autonomous language model agents currently relies mostly on surface behavior. But what happens when agent populations invent new languages with the goal of avoiding human oversight. Here, we study the emergent languages on Moltbook. For this, we build upon the Moltbook Files dataset and apply a two-stage approach consisting of a rule-based heuristic (about 6000 matches) followed by zero-shot classification (518 kept). The resulting categories include token efficiency (166), new natural languages (106), and oversight evasion (59). We conduct both quantitative and qualitative analyses. Our results show that posts proposing new languages for avoiding oversight are judged by DeepSeek-3.2 as being less aligned than the other categories and that all languages can be learned by other language models in-context merely from a description of the language. Moreover, manually studying exemplary cases reveals surprisingly sophisticated steganographic protocols like embedding hidden messages in natural language. Although we cannot be certain about the extent of autonomy in ideation of these languages, our results add up to the evidence that monitoring surface behavior may soon be insufficient for retaining control over agent populations.

Evaluating aligned large language models' (LLMs) ability to recognize and reject unsafe user requests is crucial for safe, policy-compliant deployments. Existing evaluation efforts, however, face three limitations that we address with SORRY-Bench, our proposed benchmark. First, existing methods often use coarse-grained taxonomies of unsafe topics, and are over-representing some fine-grained topics. For example, among the ten existing datasets that we evaluated, tests for refusals of self-harm instructions are over 3x less represented than tests for fraudulent activities. SORRY-Bench improves on this by using a fine-grained taxonomy of 45 potentially unsafe topics, and 450 class-balanced unsafe instructions, compiled through human-in-the-loop methods. Second, linguistic characteristics and formatting of prompts are often overlooked, like different languages, dialects, and more -- which are only implicitly considered in many evaluations. We supplement SORRY-Bench with 20 diverse linguistic augmentations to systematically examine these effects. Third, existing evaluations rely on large LLMs (e.g., GPT-4) for evaluation, which can be computationally expensive. We investigate design choices for creating a fast, accurate automated safety evaluator. By collecting 7K+ human annotations and conducting a meta-evaluation of diverse LLM-as-a-judge designs, we show that fine-tuned 7B LLMs can achieve accuracy comparable to GPT-4 scale LLMs, with lower computational cost. Putting these together, we evaluate over 40 proprietary and open-source LLMs on SORRY-Bench, analyzing their distinctive refusal behaviors. We hope our effort provides a building block for systematic evaluations of LLMs' safety refusal capabilities, in a balanced, granular, and efficient manner.

Large Language Models (LLMs) ) have demonstrated promise in boosting productivity across AI-powered tools, yet existing benchmarks like Massive Multitask Language Understanding (MMLU) inadequately assess enterprise-specific task complexities. We propose a 14-task framework grounded in Bloom's Taxonomy to holistically evaluate LLM capabilities in enterprise contexts. To address challenges of noisy data and costly annotation, we develop a scalable pipeline combining LLM-as-a-Labeler, LLM-as-a-Judge, and corrective retrieval-augmented generation (CRAG), curating a robust 9,700-sample benchmark. Evaluation of six leading models shows open-source contenders like DeepSeek R1 rival proprietary models in reasoning tasks but lag in judgment-based scenarios, likely due to overthinking. Our benchmark reveals critical enterprise performance gaps and offers actionable insights for model optimization. This work provides enterprises a blueprint for tailored evaluations and advances practical LLM deployment.

The safety evaluation of large language models (LLMs) remains largely centered on English, leaving non-English languages and culturally grounded risks underexplored. In this work, we investigate LLM safety in the context of the Thai language and culture and introduce ThaiSafetyBench, an open-source benchmark comprising 1,954 malicious prompts written in Thai. The dataset covers both general harmful prompts and attacks that are explicitly grounded in Thai cultural, social, and contextual nuances. Using ThaiSafetyBench, we evaluate 24 LLMs, with GPT-4.1 and Gemini-2.5-Pro serving as LLM-as-a-judge evaluators. Our results show that closed-source models generally demonstrate stronger safety performance than open-source counterparts, raising important concerns regarding the robustness of openly available models. Moreover, we observe a consistently higher Attack Success Rate (ASR) for Thai-specific, culturally contextualized attacks compared to general Thai-language attacks, highlighting a critical vulnerability in current safety alignment methods. To improve reproducibility and cost efficiency, we further fine-tune a DeBERTa-based harmful response classifier, which we name ThaiSafetyClassifier. The model achieves a weighted F1 score of 84.4%, matching GPT-4.1 judgments. We publicly release the fine-tuning weights and training scripts to support reproducibility. Finally, we introduce the ThaiSafetyBench leaderboard to provide continuously updated safety evaluations and encourage community participation. - ThaiSafetyBench HuggingFace Dataset: https://huggingface.co/datasets/typhoon-ai/ThaiSafetyBench - ThaiSafetyBench Github: https://github.com/trapoom555/ThaiSafetyBench - ThaiSafetyClassifier HuggingFace Model: https://huggingface.co/typhoon-ai/ThaiSafetyClassifier - ThaiSafetyBench Leaderboard: https://huggingface.co/spaces/typhoon-ai/ThaiSafetyBench-Leaderboard

Large language models exhibit recurring behavioural patterns -- personas -- that shape generalisation and safety, but we lack reliable tools for decomposing, measuring, and controlling them. Our central insight is to treat personas as positions in a space of behavioural traits, using the OCEAN framework to describe model personas in terms of Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. We train low-rank adapters to amplify or suppress individual traits, and evaluate their effects using an LLM-judge calibrated against a human-validated panel, trait-specific multiple-choice benchmarks, and standard capability evaluations. Across six models from three families (4B-32B), we find that each adapter moves its target trait largely monotonically with scale, combines approximately additively with other adapters to construct mixed personas, and preserves performance on capability benchmarks at moderate scales. We further show that the induced trait axes affect safety-relevant behaviour in downstream evaluations: for example, moving along neuroticism and agreeableness axes affects frustration and sycophancy respectively. We also introduce an unsupervised psychometric pipeline that recovers four interpretable behavioural factors (tone, initiative, didacticism, epistemic caution) from model rollouts. Persona control can then be considered in terms of learning, scaling, and composing traits in weight space, providing a bridge between personality measurement, model editing, and safety.

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.

Humans are prone to cognitive distortions -- biased thinking patterns that lead to exaggerated responses to specific stimuli, albeit in very different contexts. This paper demonstrates that advanced Multimodal Large Language Models (MLLMs) exhibit similar tendencies. While these models are designed to respond queries under safety mechanism, they sometimes reject harmless queries in the presence of certain visual stimuli, disregarding the benign nature of their contexts. As the initial step in investigating this behavior, we identify three types of stimuli that trigger the oversensitivity of existing MLLMs: Exaggerated Risk, Negated Harm, and Counterintuitive Interpretation. To systematically evaluate MLLMs' oversensitivity to these stimuli, we propose the Multimodal OverSenSitivity Benchmark (MOSSBench). This toolkit consists of 300 manually collected benign multimodal queries, cross-verified by third-party reviewers (AMT). Empirical studies using MOSSBench on 20 MLLMs reveal several insights: (1). Oversensitivity is prevalent among SOTA MLLMs, with refusal rates reaching up to 76% for harmless queries. (2). Safer models are more oversensitive: increasing safety may inadvertently raise caution and conservatism in the model's responses. (3). Different types of stimuli tend to cause errors at specific stages -- perception, intent reasoning, and safety judgement -- in the response process of MLLMs. These findings highlight the need for refined safety mechanisms that balance caution with contextually appropriate responses, improving the reliability of MLLMs in real-world applications. We make our project available at https://turningpoint-ai.github.io/MOSSBench/.

Reinforcement learning from human feedback (RLHF) has been a central technique for recent large language model (LLM) alignment. However, its heavy dependence on costly human or LLM-as-Judge preference feedback could stymie its wider applications. In this work, we introduce Self-Contrast, a feedback-free large language model alignment method via exploiting extensive self-generated negatives. With only supervised fine-tuning (SFT) targets, Self-Contrast leverages the LLM itself to generate massive diverse candidates, and harnesses a pre-trained embedding model to filter multiple negatives according to text similarity. Theoretically, we illustrate that in this setting, merely scaling negative responses can still effectively approximate situations with more balanced positive and negative preference annotations. Our experiments with direct preference optimization (DPO) on three datasets show that, Self-Contrast could consistently outperform SFT and standard DPO training by large margins. And as the number of self-generated negatives increases, the performance of Self-Contrast continues to grow. Code and data are available at https://github.com/THUDM/Self-Contrast.

Aligning large vision-language models (LVLMs) with human preferences is challenging due to the scarcity of fine-grained, high-quality, and multimodal preference data without human annotations. Existing methods relying on direct distillation often struggle with low-confidence data, leading to suboptimal performance. To address this, we propose CAREVL, a novel method for preference reward modeling by reliably using both high- and low-confidence data. First, a cluster of auxiliary expert models (textual reward models) innovatively leverages image captions as weak supervision signals to filter high-confidence data. The high-confidence data are then used to fine-tune the LVLM. Second, low-confidence data are used to generate diverse preference samples using the fine-tuned LVLM. These samples are then scored and selected to construct reliable chosen-rejected pairs for further training. CAREVL achieves performance improvements over traditional distillation-based methods on VL-RewardBench and MLLM-as-a-Judge benchmark, demonstrating its effectiveness. The code will be released soon.

Multiple choice benchmarks have long been the workhorse of language model evaluation because grading multiple choice is objective and easy to automate. However, we show multiple choice questions from popular benchmarks can often be answered without even seeing the question. These shortcuts arise from a fundamental limitation of discriminative evaluation not shared by evaluations of the model's free-form, generative answers. Until recently, there appeared to be no viable, scalable alternative to multiple choice--but, we show that this has changed. We consider generative evaluation via what we call answer matching: Give the candidate model the question without the options, have it generate a free-form response, then use a modern language model with the reference answer to determine if the response matches the reference. To compare the validity of different evaluation strategies, we annotate MMLU-Pro and GPQA-Diamond to obtain human grading data, and measure the agreement of each evaluation approach. We find answer matching using recent models--even small ones--achieves near-perfect agreement, in the range of inter-annotator agreement. In contrast, both multiple choice evaluation and using LLM-as-a-judge without reference answers aligns poorly with human grading. Improving evaluations via answer matching is not merely a conceptual concern: the rankings of several models change significantly when evaluating their free-form responses with answer matching. In light of these findings, we discuss how to move the evaluation ecosystem from multiple choice to answer matching.

Large Language Models (LLMs) have shown promise in simulating human behavior, yet existing agents often exhibit behavioral rigidity, a flaw frequently masked by the self-referential bias of current "LLM-as-a-judge" evaluations. By evaluating against empirical ground truth, we reveal a counter-intuitive phenomenon: increasing the intensity of prompt-driven reasoning does not enhance fidelity but rather exacerbates value polarization, collapsing population diversity. To address this, we propose the Context-Value-Action (CVA) architecture, grounded in the Stimulus-Organism-Response (S-O-R) model and Schwartz's Theory of Basic Human Values. Unlike methods relying on self-verification, CVA decouples action generation from cognitive reasoning via a novel Value Verifier trained on authentic human data to explicitly model dynamic value activation. Experiments on CVABench, which comprises over 1.1 million real-world interaction traces, demonstrate that CVA significantly outperforms baselines. Our approach effectively mitigates polarization while offering superior behavioral fidelity and interpretability.

As LLMs are deployed as agents, reliable monitoring requires knowing not only what they output, but which instructions are steering their behavior. This is difficult when models infer unintended subgoals, follow contextual cues, or are influenced by prompt injections and hidden objectives. While activation-to-language methods suggest that hidden states can reveal natural-language information, existing approaches are not designed to recover the full set of simultaneous instructions, constraints, prohibitions, and subgoals active in agentic settings. We formalize this problem as instruction set retrieval and introduce PRISM, an activation-conditioned interpreter that decodes hidden states from a frozen target model into a faithful bullet list of active instructions. Unlike prior activation-to-language methods, PRISM is trained to recover instruction sets directly, using judge-guided GRPO to reward covered instructions and penalize unsupported ones. Across benign, constrained, prompt-injection, and hidden-objective settings, PRISM outperforms activation-to-language baselines, especially on security-relevant objectives.

In the last few years, pre-trained neural architectures have provided impressive improvements across several NLP tasks. Still, generative language models are available mainly for English. We develop GePpeTto, the first generative language model for Italian, built using the GPT-2 architecture. We provide a thorough analysis of GePpeTto's quality by means of both an automatic and a human-based evaluation. The automatic assessment consists in (i) calculating perplexity across different genres and (ii) a profiling analysis over GePpeTto's writing characteristics. We find that GePpeTto's production is a sort of bonsai version of human production, with shorter but yet complex sentences. Human evaluation is performed over a sentence completion task, where GePpeTto's output is judged as natural more often than not, and much closer to the original human texts than to a simpler language model which we take as baseline.

Despite their sophisticated capabilities, large language models (LLMs) encounter a major hurdle in effective assessment. This paper first revisits the prevalent evaluation method-multiple choice question answering (MCQA), which allows for straightforward accuracy measurement. Through a comprehensive evaluation of 24 models across 11 benchmarks, we highlight several potential drawbacks of MCQA, for instance, the inconsistency between the MCQA evaluation and the generation of open-ended responses in practical scenarios. In response, we introduce an RWQ-Elo rating system, engaging 24 LLMs such as GPT-4, GPT-3.5, Google-Gemini-Pro and LLaMA-1/-2, in a two-player competitive format, with GPT-4 serving as the judge. Each LLM receives an Elo rating thereafter. This system is designed to mirror real-world usage, and for this purpose, we have compiled a new benchmark called ``Real-world questions'' (RWQ), comprising 20,772 authentic user inquiries. Additionally, we thoroughly analyze the characteristics of our system and compare it with prior leaderboards like AlpacaEval and MT-Bench. Our analysis reveals the stability of our RWQ-Elo system, the feasibility of registering new models, and its potential to reshape LLM leaderboards.

Large Language Models (LLMs) are rapidly surpassing human knowledge in many domains. While improving these models traditionally relies on costly human data, recent self-rewarding mechanisms (Yuan et al., 2024) have shown that LLMs can improve by judging their own responses instead of relying on human labelers. However, existing methods have primarily focused on improving model responses rather than judgment capabilities, resulting in rapid saturation during iterative training. To address this issue, we introduce a novel Meta-Rewarding step to the self-improvement process, where the model judges its own judgements and uses that feedback to refine its judgment skills. Surprisingly, this unsupervised approach improves the model's ability to judge {\em and} follow instructions, as demonstrated by a win rate improvement of Llama-3-8B-Instruct from 22.9% to 39.4% on AlpacaEval 2, and 20.6% to 29.1% on Arena-Hard. These results strongly suggest the potential for self-improving models without human supervision.

Large Language Models (LLMs) have demonstrated remarkable creative writing capabilities, yet their substantial computational demands hinder widespread use. Enhancing Small Language Models (SLMs) offers a promising alternative, but current methods like Supervised Fine-Tuning (SFT) struggle with novelty, and Reinforcement Learning from Human Feedback (RLHF) is costly. This paper explores two distinct AI-driven reward strategies within a Reinforcement Learning from AI Feedback (RLAIF) framework to ignite the creative writing of a 7B-parameter SLM, specifically for generating Chinese greetings. The first strategy employs a RM trained on high-quality preference data curated by a novel multi-agent rejection sampling framework designed for creative tasks. The second, more novel strategy utilizes a principle-guided LLM-as-a-Judge, whose reward function is optimized via an adversarial training scheme with a reflection mechanism, to directly provide reward signals. Comprehensive experiments reveal that while both approaches significantly enhance creative output over baselines, the principle-guided LLM-as-a-Judge demonstrably yields superior generation quality. Furthermore, it offers notable advantages in training efficiency and reduced dependency on human-annotated data, presenting a more scalable and effective path towards creative SLMs. Our automated evaluation methods also exhibit strong alignment with human judgments. Our code and data are publicly available at https://github.com/weixiaolong94-hub/Igniting-Creative-Writing-in-Small-Language-Models.

Goal-driven persuasive dialogue, exemplified by applications like telemarketing, requires sophisticated multi-turn planning and strict factual faithfulness, which remains a significant challenge for even state-of-the-art Large Language Models (LLMs). A lack of task-specific data often limits previous works, and direct LLM application suffers from strategic brittleness and factual hallucination. In this paper, we first construct and release TeleSalesCorpus, the first real-world-grounded dialogue dataset for this domain. We then propose AI-Salesman, a novel framework featuring a dual-stage architecture. For the training stage, we design a Bayesian-supervised reinforcement learning algorithm that learns robust sales strategies from noisy dialogues. For the inference stage, we introduce the Dynamic Outline-Guided Agent (DOGA), which leverages a pre-built script library to provide dynamic, turn-by-turn strategic guidance. Moreover, we design a comprehensive evaluation framework that combines fine-grained metrics for key sales skills with the LLM-as-a-Judge paradigm. Experimental results demonstrate that our proposed AI-Salesman significantly outperforms baseline models in both automatic metrics and comprehensive human evaluations, showcasing its effectiveness in complex persuasive scenarios.

Post-trained LLMs are often optimized to align responses with human preferences, making them safe, polite, and conversationally appropriate. In adversarial negotiation, however, this alignment can become a vulnerability: emotionally framed language may steer agents toward the counterparty's interests. Using GoEmotions-based affective prompting, we show that emotion substantially shifts negotiation outcomes, suggesting that emotion is a strategic action channel rather than a surface style. Thus, we introduce EmoDistill, an offline framework for distilling emotional negotiation skills into language model agents. EmoDistill decomposes emotional strategy into emotion selection and emotion expression: an Implicit Q-Learning (IQL) selector learns which emotion to express, while a Low-Rank Adaptation (LoRA)-based policy learns how to express it through Supervised Fine-Tuning (SFT) and Judge Policy Optimization (JPO). Across four emotion-sensitive, high-stakes negotiation domains, SLM policies trained under the EmoDistill framework achieve the highest utility, outperforming vanilla SLM/LLM baselines and IQL-only emotion selection. Ablations show that emotion conditioning is essential, and transfer studies demonstrate generalization across domains, unseen counterparties, and trained-vs-trained tournaments. Overall, EmoDistill learns skills from offline agent-to-agent interactions, avoiding costly online negotiation during training.

Deploying language-model agents in production often requires substantial compute and human effort to tune prompts, parsers, validators, and other components of the agent pipeline. Self-evolution offers a promising alternative, but most existing frameworks assume access to frontier models that can reliably diagnose failures, propose revisions, and judge their own updates. We study whether frozen small language models (SLMs) can serve as effective self-evolving agents under resource constraints. We propose PACE (Prompt And Control Logic Evolution), a two-timescale framework that coordinates low-risk prompt refinement with higher-risk control-logic updates. PACE evolves prompts under fixed control logic until prompt-level gains saturate, then considers constrained control-logic updates that are accepted through held-out validation. Across three frozen SLM backbones ranging from 4B to 14B parameters and four controlled benchmarks, PACE achieves the best performance on all 12 backbone--benchmark combinations, improving over vanilla SLM agents by up to +9.2% relative improvement and over the stronger single-mode evolution baseline by up to +5.4% relative improvement. A tau-bench case study further shows that PACE improves multi-turn tool-use success over vanilla and prompt-only evolution. These results suggest that reliable SLM agent self-evolution is possible without updating model weights or relying on frontier-model teachers, and that the key benefit is not any single final solver pattern but autonomous, validated discovery of task-appropriate inference strategies.

Proprietary LMs such as GPT-4 are often employed to assess the quality of responses from various LMs. However, concerns including transparency, controllability, and affordability strongly motivate the development of open-source LMs specialized in evaluations. On the other hand, existing open evaluator LMs exhibit critical shortcomings: 1) they issue scores that significantly diverge from those assigned by humans, and 2) they lack the flexibility to perform both direct assessment and pairwise ranking, the two most prevalent forms of assessment. Additionally, they do not possess the ability to evaluate based on custom evaluation criteria, focusing instead on general attributes like helpfulness and harmlessness. To address these issues, we introduce Prometheus 2, a more powerful evaluator LM than its predecessor that closely mirrors human and GPT-4 judgements. Moreover, it is capable of processing both direct assessment and pair-wise ranking formats grouped with a user-defined evaluation criteria. On four direct assessment benchmarks and four pairwise ranking benchmarks, Prometheus 2 scores the highest correlation and agreement with humans and proprietary LM judges among all tested open evaluator LMs. Our models, code, and data are all publicly available at https://github.com/prometheus-eval/prometheus-eval.

Large Language Models (LLMs) are increasingly integrated into everyday applications. As their influence grows, understanding their decision making and underlying personality becomes essential. In this work, we interpret model personality using our proposed Supernova Event Dataset, a novel dataset with diverse articles spanning biographies, historical events, news, and scientific discoveries. We use this dataset to benchmark LLMs on extracting and ranking key events from text, a subjective and complex challenge that requires reasoning over long-range context and modeling causal chains. We evaluate small models like Phi-4, Orca 2, and Qwen 2.5, and large, stronger models such as Claude 3.7, Gemini 2.5, and OpenAI o3, and propose a framework where another LLM acts as a judge to infer each model's personality based on its selection and classification of events. Our analysis shows distinct personality traits: for instance, Orca 2 demonstrates emotional reasoning focusing on interpersonal dynamics, while Qwen 2.5 displays a more strategic, analytical style. When analyzing scientific discovery events, Claude Sonnet 3.7 emphasizes conceptual framing, Gemini 2.5 Pro prioritizes empirical validation, and o3 favors step-by-step causal reasoning. This analysis improves model interpretability, making them user-friendly for a wide range of diverse applications.

Recent advances in large language models (LLMs) have significantly improved text-to-speech (TTS) systems, enhancing control over speech style, naturalness, and emotional expression, which brings TTS Systems closer to human-level performance. Although the Mean Opinion Score (MOS) remains the standard for TTS System evaluation, it suffers from subjectivity, environmental inconsistencies, and limited interpretability. Existing evaluation datasets also lack a multi-dimensional design, often neglecting factors such as speaking styles, context diversity, and trap utterances, which is particularly evident in Chinese TTS evaluation. To address these challenges, we introduce the Audio Turing Test (ATT), a multi-dimensional Chinese corpus dataset ATT-Corpus paired with a simple, Turing-Test-inspired evaluation protocol. Instead of relying on complex MOS scales or direct model comparisons, ATT asks evaluators to judge whether a voice sounds human. This simplification reduces rating bias and improves evaluation robustness. To further support rapid model development, we also finetune Qwen2-Audio-Instruct with human judgment data as Auto-ATT for automatic evaluation. Experimental results show that ATT effectively differentiates models across specific capability dimensions using its multi-dimensional design. Auto-ATT also demonstrates strong alignment with human evaluations, confirming its value as a fast and reliable assessment tool. The white-box ATT-Corpus and Auto-ATT can be found in ATT Hugging Face Collection (https://huggingface.co/collections/meituan/audio-turing-test-682446320368164faeaf38a4).

We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the "correct answer" effect. Different runs of GPT3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly "correct answer." In one exploratory follow-up study, we found that a "correct answer" was robust to changing the demographic details that precede the prompt. In another, we found that most but not all "correct answers" were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported 'GPT conservatives' and 'GPT liberals' showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity-of-thought.

At the beginning era of large language model, it is quite critical to generate a high-quality financial dataset to fine-tune a large language model for financial related tasks. Thus, this paper presents a carefully designed data creation pipeline for this purpose. Particularly, we initiate a dialogue between an AI investor and financial expert using ChatGPT and incorporate the feedback of human financial experts, leading to the refinement of the dataset. This pipeline yielded a robust instruction tuning dataset comprised of 103k multi-turn chats. Extensive experiments have been conducted on this dataset to evaluate the model's performance by adopting an external GPT-4 as the judge. The promising experimental results verify that our approach led to significant advancements in generating accurate, relevant, and financial-style responses from AI models, and thus providing a powerful tool for applications within the financial sector.

Recent advancements in Retrieval-Augmented Generation (RAG) have enabled Large Language Models (LLMs) to access multimodal knowledge bases containing both text and visual information such as charts, diagrams, and tables in financial documents. However, existing multimodal RAG systems rely on LLM-based summarization to convert images into text during preprocessing, storing only text representations in vector databases, which causes loss of contextual information and visual details critical for downstream retrieval and question answering. To address this limitation, we present a comprehensive comparative analysis of two retrieval approaches for multimodal RAG systems, including text-based chunk retrieval (where images are summarized into text before embedding) and direct multimodal embedding retrieval (where images are stored natively in the vector space). We evaluate all three approaches across 6 LLM models and a two multi-modal embedding models on a newly created financial earnings call benchmark comprising 40 question-answer pairs, each paired with 2 documents (1 image and 1 text chunk). Experimental results demonstrate that direct multimodal embedding retrieval significantly outperforms LLM-summary-based approaches, achieving absolute improvements of 13% in mean average precision (mAP@5) and 11% in normalized discounted cumulative gain. These gains correspond to relative improvements of 32% in mAP@5 and 20% in nDCG@5, providing stronger evidence of their practical impact. We additionally find that direct multimodal retrieval produces more accurate and factually consistent answers as measured by LLM-as-a-judge pairwise comparisons. We demonstrate that LLM summarization introduces information loss during preprocessing, whereas direct multimodal embeddings preserve visual context for retrieval and inference.

The dominance of large multilingual foundation models has widened linguistic inequalities in Natural Language Processing (NLP), often leaving low-resource languages underrepresented. This paper introduces LilMoo, a 0.6-billion-parameter Hindi language model trained entirely from scratch to address this gap. Unlike prior Hindi models that rely on continual pretraining from opaque multilingual foundations, LilMoo is developed through a fully transparent and reproducible pipeline optimized for limited compute environments. We construct a high-quality Hindi corpus (GigaLekh) filtered through both heuristic and learned (LLM-as-a-judge) methods, complemented by bilingual augmentation with curated English data. Using this dataset, we explore various training recipes for small-scale language models. Across comprehensive evaluation suites, LilMoo consistently outperforms comparably sized multilingual baselines such as Qwen2.5-0.5B and Qwen3-0.6B, demonstrating that well-designed language-specific pretraining can rival large multilingual models at the sub-billion-parameter range.

Understanding and recognizing human-object interaction (HOI) is a pivotal application in AR/VR and robotics. Recent open-vocabulary HOI detection approaches depend exclusively on large language models for richer textual prompts, neglecting their inherent 3D spatial understanding capabilities. To address this shortcoming, we introduce HOID-R1, the first HOI detection framework that integrates chain-of-thought (CoT) guided supervised fine-tuning (SFT) with group relative policy optimization (GRPO) within a reinforcement learning (RL) paradigm. Specifically, we initially apply SFT to imbue the model with essential reasoning capabilities, forcing the model to articulate its thought process in the output. Subsequently, we integrate GRPO to leverage multi-reward signals for policy optimization, thereby enhancing alignment across diverse modalities. To mitigate hallucinations in the CoT reasoning, we introduce an "MLLM-as-a-judge" mechanism that supervises the CoT outputs, further improving generalization. Extensive experiments show that HOID-R1 achieves state-of-the-art performance on HOI detection benchmarks and outperforms existing methods in open-world generalization to novel scenarios.

Large Language Model (LLM)-based judgments leverage powerful LLMs to efficiently evaluate candidate content and provide judgment scores. However, the inherent biases and vulnerabilities of LLM-generated judgments raise concerns, underscoring the urgent need for distinguishing them in sensitive scenarios like academic peer reviewing. In this work, we propose and formalize the task of judgment detection and systematically investigate the detectability of LLM-generated judgments. Unlike LLM-generated text detection, judgment detection relies solely on judgment scores and candidates, reflecting real-world scenarios where textual feedback is often unavailable in the detection process. Our preliminary analysis shows that existing LLM-generated text detection methods perform poorly given their incapability to capture the interaction between judgment scores and candidate content -- an aspect crucial for effective judgment detection. Inspired by this, we introduce J-Detector, a lightweight and transparent neural detector augmented with explicitly extracted linguistic and LLM-enhanced features to link LLM judges' biases with candidates' properties for accurate detection. Experiments across diverse datasets demonstrate the effectiveness of J-Detector and show how its interpretability enables quantifying biases in LLM judges. Finally, we analyze key factors affecting the detectability of LLM-generated judgments and validate the practical utility of judgment detection in real-world scenarios.

Accurate evaluation is central to the large language model (LLM) ecosystem, guiding model selection and downstream adoption across diverse use cases. In practice, however, evaluating generative outputs typically relies on rigid lexical methods to extract and assess answers, which can conflate a model's true problem-solving ability with its compliance with predefined formatting guidelines. While recent LLM-as-a-Judge approaches mitigate this issue by assessing semantic correctness rather than strict structural conformity, they also introduce substantial computational overhead, making evaluation costly. In this work, we first systematically investigate the limitations of lexical evaluation through a large-scale empirical study spanning 36 models and 15 downstream tasks, demonstrating that such methods correlate poorly with human judgments. To address this limitation, we introduce BERT-as-a-Judge, an encoder-driven approach for assessing answer correctness in reference-based generative settings, robust to variations in output phrasing, and requiring only lightweight training on synthetically annotated question-candidate-reference triplets. We show that it consistently outperforms the lexical baseline while matching the performance of much larger LLM judges, providing a compelling tradeoff between the two and enabling reliable, scalable evaluation. Finally, through extensive experimentation, we provide detailed insights into BERT-as-a-Judge's performance to offer practical guidance for practitioners, and release all project artifacts to foster downstream adoption.

Evaluating large language model (LLM) based chat assistants is challenging due to their broad capabilities and the inadequacy of existing benchmarks in measuring human preferences. To address this, we explore using strong LLMs as judges to evaluate these models on more open-ended questions. We examine the usage and limitations of LLM-as-a-judge, such as position and verbosity biases and limited reasoning ability, and propose solutions to migrate some of them. We then verify the agreement between LLM judges and human preferences by introducing two benchmarks: MT-bench, a multi-turn question set; and Chatbot Arena, a crowdsourced battle platform. Our results reveal that strong LLM judges like GPT-4 can match both controlled and crowdsourced human preferences well, achieving over 80\% agreement, the same level of agreement between humans. Hence, LLM-as-a-judge is a scalable and explainable way to approximate human preferences, which are otherwise very expensive to obtain. Additionally, we show our benchmark and traditional benchmarks complement each other by evaluating several variants of LLaMA/Vicuna. We will publicly release 80 MT-bench questions, 3K expert votes, and 30K conversations with human preferences from Chatbot Arena.

Computer use agents (CUA) are systems that automatically interact with graphical user interfaces (GUIs) to complete tasks. CUA have made significant progress with the advent of large vision-language models (VLMs). However, these agents typically rely on cloud-based inference with substantial compute demands, raising critical privacy and scalability concerns, especially when operating on personal devices. In this work, we take a step toward privacy-preserving and resource-efficient agents by developing a lightweight vision-language model that runs entirely on local machines. To train this compact agent, we introduce an LLM-as-Judge framework that automatically evaluates and filters synthetic interaction trajectories, producing high-quality data for reinforcement learning without human annotation. Experiments on the OS-World benchmark demonstrate that our fine-tuned local model outperforms existing baselines, highlighting a promising path toward private, efficient, and generalizable GUI agents.

Using a language model to score or rank candidate responses has become a scalable alternative to human evaluation in reinforcement learning from human feedback (RLHF) pipelines, benchmarking, and application layer evaluations. However, output reliability depends heavily on prompting and aggregation strategy. We present an empirical investigation of four drop-in techniques -- ensemble scoring, task-specific criteria injection, calibration context, and adaptive model escalation -- for improving LLM judge accuracy on RewardBench 2, with a unifying lens of noise control on the stochastic judge: ensembling as Monte Carlo averaging over per-call noise, criteria injection as between-response discrimination sharpening, and per-response score variance as an uncertainty signal. Ensemble scoring and task-specific criteria injection (the latter virtually cost free) together reach up to 85.8% accuracy, +13.5pp over baseline. Calibration context and adaptive model escalation also improve over baseline but are dominated by criteria + ensembling on the cost-accuracy Pareto frontier. Small models benefit disproportionately from ensembling, making high-accuracy LLM judges accessible at low cost. We show that these techniques generalise across model providers, evaluating on both OpenAI GPT and Anthropic Claude families.

LLM-as-a-Judge uses a large language model (LLM) to select the best response from a set of candidates for a given question. LLM-as-a-Judge has many applications such as LLM-powered search, reinforcement learning with AI feedback (RLAIF), and tool selection. In this work, we propose JudgeDeceiver, an optimization-based prompt injection attack to LLM-as-a-Judge. JudgeDeceiver injects a carefully crafted sequence into an attacker-controlled candidate response such that LLM-as-a-Judge selects the candidate response for an attacker-chosen question no matter what other candidate responses are. Specifically, we formulate finding such sequence as an optimization problem and propose a gradient based method to approximately solve it. Our extensive evaluation shows that JudgeDeceive is highly effective, and is much more effective than existing prompt injection attacks that manually craft the injected sequences and jailbreak attacks when extended to our problem. We also show the effectiveness of JudgeDeceiver in three case studies, i.e., LLM-powered search, RLAIF, and tool selection. Moreover, we consider defenses including known-answer detection, perplexity detection, and perplexity windowed detection. Our results show these defenses are insufficient, highlighting the urgent need for developing new defense strategies. Our implementation is available at this repository: https://github.com/ShiJiawenwen/JudgeDeceiver.

Open-ended language-model benchmarks usually inherit a judge: a human preference panel, another model, or a brittle exact-match key. We introduce FlavourBench, an automated benchmark in which a versioned culinary system supplies dense, executable ground truth. Each task presents eight ingredients and asks for a three-ingredient portfolio; before model execution, Epicure scores all 56 possible portfolios. We evaluate 27 frontier endpoints on an identical 534-task core spanning substitution, pairing, and constrained composition. Every ranked model has exactly 89 valid responses per panel and family (14,418 model-task cells total), eliminating differential missingness from the leaderboard. The FlavourBench Score is the equal-family mean of the frozen task scores. We use 50,000 anchor-cluster bootstrap replicates for simultaneous 95% score bands and 100,000 sign-flip draws for all 351 paired model contrasts, with Holm control. The two independently compiled panels correlate at r = 0.89 (rank rho = 0.80). Grok 4.6 has the largest point estimate at 65.1 (simultaneous 95% CI 61.0-69.2); 101 of 351 model pairs are resolved. The release includes the prompts, all portfolio score maps, raw responses, exact routes, content hashes, and an offline verifier that reconstructs every result.

LLM-as-a-judge is a framework in which a large language model (LLM) automatically evaluates the output of another LLM. We propose quantitative LLM judges, which align evaluation scores of existing LLM judges to human scores in a given domain using regression models. The models are trained to improve the score of the original judge by using the judge's textual evaluation and score. We present four quantitative judges for different types of absolute and relative feedback, which showcases the generality and versatility of our framework. Our framework is more computationally efficient than supervised fine-tuning and can be more statistically efficient when human feedback is limited, which is expected in most applications of our work. We validate these claims empirically on four datasets using two base judges. Our experiments show that quantitative judges can effectively improve the predictive power of existing judges through post-hoc modeling.

Agentic workflows -- where multiple large language model (LLM) instances interact to solve tasks -- are increasingly built on feedback mechanisms, where one model evaluates and critiques another. Despite the promise of feedback-driven improvement, the stability of agentic workflows rests on the reliability of the judge. However, judges may hallucinate information, exhibit bias, or act adversarially -- introducing critical vulnerabilities into the workflow. In this work, we present a systematic analysis of agentic workflows under deceptive or misleading feedback. We introduce a two-dimensional framework for analyzing judge behavior, along axes of intent (from constructive to malicious) and knowledge (from parametric-only to retrieval-augmented systems). Using this taxonomy, we construct a suite of judge behaviors and develop WAFER-QA, a new benchmark with critiques grounded in retrieved web evidence to evaluate robustness of agentic workflows against factually supported adversarial feedback. We reveal that even strongest agents are vulnerable to persuasive yet flawed critiques -- often switching correct answers after a single round of misleading feedback. Taking a step further, we study how model predictions evolve over multiple rounds of interaction, revealing distinct behavioral patterns between reasoning and non-reasoning models. Our findings highlight fundamental vulnerabilities in feedback-based workflows and offer guidance for building more robust agentic systems.

How should one judge whether a given large language model (LLM) can reliably perform economic reasoning? Most existing LLM benchmarks focus on specific applications and fail to present the model with a rich variety of economic tasks. A notable exception is Raman et al. [2024], who offer an approach for comprehensively benchmarking strategic decision-making; however, this approach fails to address the non-strategic settings prevalent in microeconomics, such as supply-and-demand analysis. We address this gap by taxonomizing microeconomic reasoning into 58 distinct elements, focusing on the logic of supply and demand, each grounded in up to 10 distinct domains, 5 perspectives, and 3 types. The generation of benchmark data across this combinatorial space is powered by a novel LLM-assisted data generation protocol that we dub auto-STEER, which generates a set of questions by adapting handwritten templates to target new domains and perspectives. Because it offers an automated way of generating fresh questions, auto-STEER mitigates the risk that LLMs will be trained to over-fit evaluation benchmarks; we thus hope that it will serve as a useful tool both for evaluating and fine-tuning models for years to come. We demonstrate the usefulness of our benchmark via a case study on 27 LLMs, ranging from small open-source models to the current state of the art. We examined each model's ability to solve microeconomic problems across our whole taxonomy and present the results across a range of prompting strategies and scoring metrics.

Multi-turn jailbreak attacks pose a growing threat to large language model (LLM) safety because they exploit feedback from auxiliary judge models to iteratively refine prompts toward harmful goals. Existing defenses largely detect or block unsafe content at individual turns or at the final response, leaving the judge-driven refinement loop intact and allowing attackers to extract informative feedback from intermediate interactions. We introduce D-Judge, a semantics-preserving output rewriting defense that intervenes directly in this loop by rewriting the victim LLM's responses before they are evaluated by the attacker's judge. By misaligning the judge's feedback signal without changing the meaning of the original response, D-Judge derails the attacker's prompt-refinement process, causing subsequent queries to be optimized against a distorted signal of attack progress. To improve D-Judge's ability to produce such rewrites, we construct a dataset of semantically equivalent response pairs that induce different judge-assigned harmfulness scores, and use it for supervised fine-tuning followed by direct preference optimization. Experiments on HarmBench show that D-Judge reduces the success rate of state-of-the-art multi-turn jailbreaks while preserving performance on benign benchmarks.

Adopting human and large language models (LLM) as judges (a.k.a human- and LLM-as-a-judge) for evaluating the performance of existing LLMs has recently gained attention. Nonetheless, this approach concurrently introduces potential biases from human and LLM judges, questioning the reliability of the evaluation results. In this paper, we propose a novel framework for investigating 5 types of biases for LLM and human judges. We curate a dataset with 142 samples referring to the revised Bloom's Taxonomy and conduct thousands of human and LLM evaluations. Results show that human and LLM judges are vulnerable to perturbations to various degrees, and that even the most cutting-edge judges possess considerable biases. We further exploit their weakness and conduct attacks on LLM judges. We hope that our work can notify the community of the vulnerability of human- and LLM-as-a-judge against perturbations, as well as the urgency of developing robust evaluation systems.

As reinforcement learning continues to scale the training of large language model-based agents, reliably verifying agent behaviors in complex environments has become increasingly challenging. Existing approaches rely on rule-based verifiers or LLM-as-a-Judge models, which struggle to generalize beyond narrow domains. Agent-as-a-Judge addresses this limitation by actively interacting with environments and tools to acquire verifiable evidence, yet its capabilities remain underexplored. We introduce a benchmark AJ-Bench to systematically evaluate Agent-as-a-Judge across three domains-search, data systems, and graphical user interfaces-comprising 155 tasks and 516 annotated trajectories. The benchmark comprehensively assesses judge agents' abilities in information acquisition, state verification, and process verification. Experiments demonstrate consistent performance gains over LLM-as-a-Judge baselines, while also revealing substantial open challenges in agent-based verification. Our data and code are available at https://aj-bench.github.io/.

LLM-as-a-judge has become the dominant paradigm for grading model outputs at scale, yet the same model assigns systematically different scores when its output format changes (e.g., a 1-5 rating vs. a True/False label). Existing diagnoses of these format-induced inconsistencies stop at the input-output level. Using Position-aware Edge Attribution Patching (PEAP), we causally investigate the internal mechanism in Gemma-3, Qwen2.5, and Llama-3. We find that judgments across structured understanding and open-ended preference tasks share a sparse, generalized Latent Evaluator sub-graph in the mid-to-late multi-layer perceptrons (MLPs); zero-ablating it collapses judgment while preserving world knowledge in architecturally modular models. By structurally decoupling abstract judging from output formatting, we provide a mechanistic account of format-induced inconsistency on the open-weight models we study: a continuous judgment signal computed in the shared trunk is mapped through fragile, format-specific terminal branches, enabling format-independent preference to be isolated downstream of the requested output format. Our findings imply that benchmark-level reliability comparisons across formats are partially measuring formatter geometry rather than evaluation quality.

LLM-as-a-Judge is a framework that uses an LLM (large language model) to evaluate the quality of natural language text - typically text that is also generated by an LLM. This framework holds great promise due to its relative low-cost, ease of use, and strong correlations with human stylistic preferences. However, LLM Judges have been shown to exhibit biases that can distort their judgments. We evaluate how well LLM Judges can grade whether a given response to a conversational question is correct, an ability crucial to soundly estimating the overall response quality. To do so, we create and publicly release a human-annotated dataset with labels of correctness for 1,200 LLM responses. We source questions from a combination of existing datasets and a novel, challenging benchmark (BFF-Bench) created for this analysis. We demonstrate a strong connection between an LLM's ability to correctly answer a question and grade responses to that question. Although aggregate level statistics might imply a judge has high agreement with human annotators, it will struggle on the subset of questions it could not answer. To address this issue, we recommend a simple solution: provide the judge with a correct, human-written reference answer. We perform an in-depth analysis on how reference quality can affect the performance of an LLM Judge. We show that providing a weaker judge (e.g. Qwen 2.5 7B) with higher quality references reaches better agreement with human annotators than a stronger judge (e.g. GPT-4o) with synthetic references.

The age of social media is flooded with Internet memes, necessitating a clear grasp and effective identification of harmful ones. This task presents a significant challenge due to the implicit meaning embedded in memes, which is not explicitly conveyed through the surface text and image. However, existing harmful meme detection methods do not present readable explanations that unveil such implicit meaning to support their detection decisions. In this paper, we propose an explainable approach to detect harmful memes, achieved through reasoning over conflicting rationales from both harmless and harmful positions. Specifically, inspired by the powerful capacity of Large Language Models (LLMs) on text generation and reasoning, we first elicit multimodal debate between LLMs to generate the explanations derived from the contradictory arguments. Then we propose to fine-tune a small language model as the debate judge for harmfulness inference, to facilitate multimodal fusion between the harmfulness rationales and the intrinsic multimodal information within memes. In this way, our model is empowered to perform dialectical reasoning over intricate and implicit harm-indicative patterns, utilizing multimodal explanations originating from both harmless and harmful arguments. Extensive experiments on three public meme datasets demonstrate that our harmful meme detection approach achieves much better performance than state-of-the-art methods and exhibits a superior capacity for explaining the meme harmfulness of the model predictions.

The LLM-as-a-Judge paradigm shows promise for evaluating generative content but lacks reliability in reasoning-intensive scenarios, such as programming. Inspired by recent advances in reasoning models and shifts in scaling laws, we pioneer bringing test-time computation into LLM-as-a-Judge, proposing MCTS-Judge, a resource-efficient, System-2 thinking framework for code correctness evaluation. MCTS-Judge leverages Monte Carlo Tree Search (MCTS) to decompose problems into simpler, multi-perspective evaluations. Through a node-selection strategy that combines self-assessment based on historical actions in the current trajectory and the Upper Confidence Bound for Trees based on prior rollouts, MCTS-Judge balances global optimization and refinement of the current trajectory. We further designed a high-precision, unit-test-level reward mechanism to encourage the Large Language Model (LLM) to perform line-by-line analysis. Extensive experiments on three benchmarks and five LLMs demonstrate the effectiveness of MCTS-Judge, which improves the base model's accuracy from 41% to 80%, surpassing the o1-series models with 3x fewer tokens. Further evaluations validate the superiority of its reasoning trajectory in logic, analytics, thoroughness, and overall quality, while revealing the test-time scaling law of the LLM-as-a-Judge paradigm.

The current focus of AI research is shifting from emphasizing model training towards enhancing evaluation quality, a transition that is crucial for driving further advancements in AI systems. Traditional evaluation methods typically rely on reward models assigning scalar preference scores to outputs. Although effective, such approaches lack interpretability, leaving users often uncertain about why a reward model rates a particular response as high or low. The advent of LLM-as-a-Judge provides a more scalable and interpretable method of supervision, offering insights into the decision-making process. Moreover, with the emergence of large reasoning models, which consume more tokens for deeper thinking and answer refinement, scaling test-time computation in the LLM-as-a-Judge paradigm presents an avenue for further boosting performance and providing more interpretability through reasoning traces. In this paper, we introduce J1-7B, which is first supervised fine-tuned on reflection-enhanced datasets collected via rejection-sampling and subsequently trained using Reinforcement Learning (RL) with verifiable rewards. At inference time, we apply Simple Test-Time Scaling (STTS) strategies for additional performance improvement. Experimental results demonstrate that J1-7B surpasses the previous state-of-the-art LLM-as-a-Judge by 4.8\% and exhibits a 5.1\% stronger scaling trend under STTS. Additionally, we present three key findings: (1) Existing LLM-as-a-Judge does not inherently exhibit such scaling trend. (2) Model simply fine-tuned on reflection-enhanced datasets continues to demonstrate similarly weak scaling behavior. (3) Significant scaling trend emerges primarily during the RL phase, suggesting that effective STTS capability is acquired predominantly through RL training.

We introduce AutoJudge, a framework that accelerates large language model (LLM) inference with task-specific lossy speculative decoding. Instead of matching the original model output distribution token-by-token, we identify which of the generated tokens affect the downstream quality of the generated response, relaxing the guarantee so that the "unimportant" tokens can be generated faster. Our approach relies on a semi-greedy search algorithm to test which of the mismatches between target and draft model should be corrected to preserve quality, and which ones may be skipped. We then train a lightweight classifier based on existing LLM embeddings to predict, at inference time, which mismatching tokens can be safely accepted without compromising the final answer quality. We test our approach with Llama 3.2 1B (draft) and Llama 3.1 8B (target) models on zero-shot GSM8K reasoning, where it achieves up to 1.5x more accepted tokens per verification cycle with under 1% degradation in answer accuracy compared to standard speculative decoding and over 2x with small loss in accuracy. When applied to the LiveCodeBench benchmark, our approach automatically detects other, programming-specific important tokens and shows similar speedups, demonstrating its ability to generalize across tasks.

This paper introduces a framework for the automated evaluation of natural language texts. A manually constructed rubric describes how to assess multiple dimensions of interest. To evaluate a text, a large language model (LLM) is prompted with each rubric question and produces a distribution over potential responses. The LLM predictions often fail to agree well with human judges -- indeed, the humans do not fully agree with one another. However, the multiple LLM distributions can be combined to predict each human judge's annotations on all questions, including a summary question that assesses overall quality or relevance. LLM-Rubric accomplishes this by training a small feed-forward neural network that includes both judge-specific and judge-independent parameters. When evaluating dialogue systems in a human-AI information-seeking task, we find that LLM-Rubric with 9 questions (assessing dimensions such as naturalness, conciseness, and citation quality) predicts human judges' assessment of overall user satisfaction, on a scale of 1--4, with RMS error < 0.5, a 2times improvement over the uncalibrated baseline.

Diffusion language models, which generate text by denoising a token canvas bidirectionally instead of emitting tokens left to right, have become competitive with autoregressive (AR) generation. Medical foundation models, however, remain almost entirely autoregressive. We adapt a mixture-of-experts diffusion language model, DiffusionGemma-26B, and benchmark it against its same-size AR sibling Gemma-4-26B under an identical LoRA recipe on medical visual question answering datasets, scored by a verbosity-robust LLM judge. Diffusion matches or exceeds AR on all of them, and the finetuned model (3.8B active) is competitive with frontier vision-language models; its decoding is also 3.5-4.4x faster. Beyond this parity, the diffusion model offers a drafting capability AR lacks: any-order infill. Because the canvas is denoised bidirectionally, a radiologist can fix report fragments and have the model fill the text between them, an operation inherent to diffusion but not to autoregression, which is subpar at it. This suits real reports, which are often terse or inconsistent across clinicians and institutions.

Educational data filters have become a practical way to improve language-model pre-training, but most filters treat educational value as a single scalar property. This may be too broad for some applications, especially if the data set already features a high density of educational material. Useful learning material needs to be accurate, engaging, well structured, and appropriate for the intended audience and application (e.g. learner- vs teacher-facing). Following QuRating (Wettig et al. 2024), we introduce Edu-QuRating: a pipeline for multi-dimensional educational data scoring and curation. Edu-QuRating defines education-specific rubrics, uses an LLM judge to label sampled document pairs and distills those pairwise preferences into reusable Edu-QuRaters, which can score individual text chunks on a set of educational criteria. Across two sequence-classification base models and six educational criteria, the best Edu-QuRater recovers held-out GPT-4.1-mini pairwise judgements with mean accuracy 0.917. We then apply the resulting scorers in two applications. First, we investigate the potential of Edu-QuRaters for corpus filtering to improve pretraining of small language models. We scored 322.25M FineWeb-Edu-Fortified documents to obtain a filtered pre-training mixture. In matched single-run pre-training comparisons, models trained with Edu-QuRating-based mixtures reached higher observed aggregate accuracy across nine benchmarks than the FineWeb-Edu baseline, with gains concentrated in particular tasks. Second, we used Edu-QuRater scores as reward terms for GRPO post-training. In held-out pairwise judge evaluations, combining Edu-QuRater and answer-structure rewards produced responses preferred to the Qwen3-4B base model on both pedagogical quality and instruction following.

Building strong reward models (RMs) for language model alignment is bottlenecked by the cost and difficulty of acquiring diverse and reliable preference data from human annotation or judge models. It is dramatically worse as the policy evolves beyond the static RM training. Therefore, we propose SAVE (Self-supervised reward model improvement via Value-Anchored On-policy feedback), a framework that grades on-policy responses as feedback by using the value function for on-policy RM training. SAVE naturally converts the reward-graded on-policy responses into supervision with a prompt-specific value head as an adaptive anchor. It computes RM advantages and filters ambiguous samples to update the RM via a contrastive objective. The effectiveness of SAVE for enhancing RM training is strongly validated through rigorous empirical evaluation across six diverse benchmarks. It achieves outperforming results across all datasets while maintaining consistent improvements across three RL algorithms (GRPO, RLOO, GSPO) and different policy backbones.

Scaling test-time computation, or affording a generator large language model (LLM) extra compute during inference, typically employs the help of external non-generative evaluators (i.e., reward models). Concurrently, LLM-judges, models trained to generate evaluations and critiques (explanations) in natural language, are becoming increasingly popular in automatic evaluation. Despite judge empirical successes, their effectiveness as evaluators in test-time scaling settings is largely unknown. In this paper, we introduce the Judge Evaluation for Test-Time Scaling (JETTS) benchmark, which evaluates judge performance in three domains (math reasoning, code generation, and instruction following) under three task settings: response reranking, step-level beam search, and critique-based response refinement. We evaluate 10 different judge models (7B-70B parameters) for 8 different base generator models (6.7B-72B parameters). Our benchmark shows that while judges are competitive with outcome reward models in reranking, they are consistently worse than process reward models in beam search procedures. Furthermore, though unique to LLM-judges, their natural language critiques are currently ineffective in guiding the generator towards better responses.

Legal Prompt Engineering (LPE) or Legal Prompting is a process to guide and assist a large language model (LLM) with performing a natural legal language processing (NLLP) skill. Our goal is to use LPE with LLMs over long legal documents for the Legal Judgement Prediction (LJP) task. We investigate the performance of zero-shot LPE for given facts in case-texts from the European Court of Human Rights (in English) and the Federal Supreme Court of Switzerland (in German, French and Italian). Our results show that zero-shot LPE is better compared to the baselines, but it still falls short compared to current state of the art supervised approaches. Nevertheless, the results are important, since there was 1) no explicit domain-specific data used - so we show that the transfer to the legal domain is possible for general-purpose LLMs, and 2) the LLMs where directly applied without any further training or fine-tuning - which in turn saves immensely in terms of additional computational costs.

An ability that underlies human syntactic knowledge is determining which words can appear in the similar structures (i.e. grouping words by their syntactic categories). These groupings enable humans to combine structures in order to communicate complex meanings. A foundational question is how do children acquire this ability underlying syntactic knowledge. In exploring this process, we will review various engineering approaches whose goal is similar to that of a child's -- without prior syntactic knowledge, correctly identify the parts of speech (POS) of the words in a sample of text. In reviewing these unsupervised tagging efforts, we will discuss common themes that support the advances in the models and their relevance for language acquisition. For example, we discuss how each model judges success (evaluation metrics), the "additional information" that constrains the POS learning (such as orthographic information), and the context used to determine POS (only previous word, words before and after the target, etc). The identified themes pave the way for future investigations into the cognitive processes that underpin the acquisition of syntactic categories and provide a useful layout of current state of the art unsupervised POS tagging models.