搜索结果：GPT-55

Objective: To describe the design and early clinical evaluation of The Daily Dose (TDD), an LLM-driven, automated clinical summarization and clinical-trial identification system integrated into routine radiation oncology practice. Design: Mixed-methods evaluation using a cross-sectional, anonymous clinician survey administered after 1 month of system deployment. Exposure: Daily automated delivery of physician-specific email summaries generated using RadOnc-GPT, including patient schedules, concise EHR-derived clinical-status summaries, and automated identification of potentially relevant clinical trials for new or consult visits. Main Outcomes and Measures: Primary outcomes included self-reported usability, satisfaction, perceived usefulness, perceived impact on workflow, time savings, and intention for continued use. Internal consistency reliability was assessed using Cronbach's $α$. Results: Among 55 respondents, 52 (94.5\%) worked in radiation oncology, and 38 (69.1\%) were attending physicians. Most participants (83.6\%) reported using TDD daily or several times per week. Mean (SD) scores were 3.89 (1.04) for usability and satisfaction, 3.43 (1.24) for perceived usefulness, and

While skill optimization for autonomous agents has gained traction, existing methods rely on complex pipelines. This leaves a fundamental question unaddressed: What constitutes a minimal viable pipeline for skill optimization, where every component is justified by theory or empirical necessity? We formalize skill optimization via Zeroth-Order (ZO) optimization, mapping classical counterparts (central difference, trust regions) to recent literature. Noting that unlike blind numerical perturbations in classical ZO, skill trajectories serve as interpretable debugging feedback. Grounded in Claude Code philosophy and PAC learning, we establish three principles for convergence and generalization: file-system-based trajectory exploration, consensus attribute mining, and independent validation gating. Eliminating redundancies, we propose SkillOpt-Lite. It accelerates convergence and outperforms full SkillOpt: improving LiveMath by +8.8 points on GPT-5.5 and +25.4 points on GPT-5.4-nano, allowing the nano model to surpass standard GPT-5.4 optimized by SkillOpt. Finally, we integrate our framework into production coding agents like VSCode Copilot, enabling developers to evolve agent skills v

Large Language Models (LLMs) can be helpful for literature search and summarisation, but retracted articles can confuse them. This article asks three open weights (offline) LLMs whether 161 high profile retracted articles had been retracted, performing a similar check for a benchmark multidisciplinary set of 34,070 non-retracted articles. Based on titles and abstracts, in over 80% of cases the LLMs claimed that a retracted article had not been retracted (GPT OSS 120B: 82%; Gemma 3 27B: 84%; DeepSeek R1 72B: 88%). The reasons given for a correct retraction declaration were often wrong, even if detailed. This confirms that LLMs have little ability to distinguish between valid and retracted studies, unless they are allowed to, and do, check online. For the benchmark test, there were only 55 false retraction claims from 34,070 non-retracted full text articles, and 28 false claims when only the title and abstract were entered, suggesting that there is only a small chance that LLMs discount valid studies. When retractions are erroneously claimed, this does not seem to be due to mistakes in the article. Overall, the results give new reasons to be cautious about LLM claims about academic f

AI agents are increasingly useful for biological data analysis, but existing benchmarks mostly test broad biological knowledge, executable workflows, or localized analysis steps rather than end-to-end scientific reasoning over spatial measurements. We introduce SpatialBench-Long, a benchmark for long-horizon spatial biology in which agents must recover biological claims from raw or near-raw data and calibrated experimental context without prescribed methods. SpatialBench-Long contains 24 evaluations across primary pancreatic ductal adenocarcinoma (PDAC), engineered glioblastoma organoids and in vivo tumors, Cas9 lineage-traced lung adenocarcinoma, and mouse optic nerve aging/intervention systems, spanning CosMx, Visium, Xenium, multiplexed error-robust fluorescence in situ hybridization (MERFISH), single-cell RNA sequencing (scRNA-seq), Slide-seq, Slide-tags, histology, and lineage-recording data. Candidate claims are hardened through reproduction, independent scientist review, and trajectory inspection. Final answers are graded deterministically over controlled vocabularies and symbols with companion rubrics capturing progress through key analysis chokepoints. Across the SpatialBe

Run the same LLM agent on the same task twice: do you get the same behavior? We find the answer is often no. In a study of 3,000 agent runs across three models (Llama 3.1 70B, GPT-4o, and Claude Sonnet 4.5) on HotpotQA, we observe that ReAct-style agents produce 2.0--4.2 distinct action sequences per 10 runs on average, even with identical inputs. More importantly, this variance predicts failure: tasks with consistent behavior ($\leq$2 unique paths) achieve 80--92% accuracy, while highly inconsistent tasks ($\geq$6 unique paths) achieve only 25--60%, a 32--55 percentage point gap depending on model. We trace variance to early decisions: 69% of divergence occurs at step 2, the first search query. Our results suggest that monitoring behavioral consistency during execution could enable early error detection and improve agent reliability.

The misuse of Java security APIs is a serious security problem in software development. Research in 2024 has shown that this problem is widespread in LLM-generated code. However, it remains unclear whether this phenomenon persists in current models and how external security knowledge affects it. This paper presents a scoped replication and extension of Mousavi et al.'s study on the Java Cryptography Architecture (JCA) and Java Secure Socket Extension (JSSE) APIs. We focus on two complementary settings: GPT-5.5 as a frontier proprietary coding model, and Llama-3.3-70B-Instruct as a strong open-weight model relevant to self-hosted deployment. The results show that although newer LLMs perform better in using Java security APIs, the problem of Java security API misuse has not been eliminated. External security knowledge substantially improves the measured outcome, but its effect is model-dependent. For Llama-3.3-70B-Instruct, secure code examples are the most effective single knowledge type. For GPT-5.5, explicit misuse patterns eliminate all detected security API misuses among valid programs in our benchmark, although some outputs remain invalid due to compilation errors or target-API

LLM-agent training pipelines routinely discard failed trajectories even though GPT-4o achieves only 14-20% on WebArena and below 55% pass@1 on ToolBench; even specialised systems at 50-65% leave the majority of trajectories unused. We introduce AgentHER, which recovers this lost signal by adapting Hindsight Experience Replay (HER) to natural-language agent trajectories: a trajectory that fails goal A is often a correct demonstration for an achievable alternative goal B. AgentHER realises this through a four-stage pipeline (failure classification, outcome extraction, LLM-guided relabeling with confidence gating, and data packaging) that converts discarded failures into SFT, DPO, and ShareGPT training data. On WebArena and ToolBench under a strict task-disjoint held-out protocol, AgentHER improves over success-only SFT by +7.6-11.4% across four model families (GPT-4o, Qwen2.5-72B/7B, LLaMA-3.1-8B), achieves 2x sample efficiency, and beats the strongest experience-centric baseline (Agent Workflow Memory) by +3.0-6.2%. Two robustness mechanisms, failure-severity weighting and cross-model multi-judge verification (gpt-4o-mini paired with Qwen2.5-72B-Instruct), reduce label noise from 5.

Deploying autonomous edge robotics in dynamic military environments is constrained by both scarce domain-specific training data and the computational limits of edge hardware. This paper introduces a hierarchical, zero-shot framework that cascades lightweight object detection with compact Vision-Language Models (VLMs) from the Qwen and Gemma families (4B-12B parameters). Grounding DINO serves as a high-recall, text-promptable region proposer, and frames with high detection confidence are passed to edge-class VLMs for semantic verification. We evaluate this pipeline on 55 high-fidelity synthetic videos from Battlefield 6 across three tasks: false-positive filtering (up to 100% accuracy), damage assessment (up to 97.5%), and fine-grained vehicle classification (55-90%). We further extend the pipeline into an agentic Scout-Commander workflow, achieving 100% correct asset deployment and a 9.8/10 reasoning score (graded by GPT-4o) with sub-75-second latency. A novel "Controlled Input" methodology decouples perception from reasoning, revealing distinct failure phenotypes: Gemma3-12B excels at tactical logic but fails in visual perception, while Gemma3-4B exhibits reasoning collapse even w

Search-engine date filters are widely used to enforce pre-cutoff retrieval in retrospective evaluations of search-augmented forecasters. We show this approach is unreliable across two major search engines: auditing Google Search's before: filter and DuckDuckGo's date-range filter, we find that at least one retrieved page contains major post-cutoff leakage for 71% of questions on Google and 81% on DuckDuckGo, and the answer is directly revealed for 41% and 55%, respectively. Using gpt-oss-120b to forecast with these leaky documents, we demonstrate inflated prediction accuracy (Brier score 0.10 vs. 0.24 with leak-free documents). We characterize recurring leakage mechanisms, including updated articles, related-content modules, unreliable metadata, and absence-based signals, and argue that date-restricted search on these engines is insufficient for credible retrospective evaluation. We recommend stronger retrieval safeguards or evaluation on frozen, time-stamped web snapshots.

The Engineering Reasoning and Instruction (ERI) benchmark is a taxonomy-driven instruction dataset designed to train and evaluate engineering-capable large language models (LLMs) and agents. This dataset spans nine engineering fields (namely: civil, mechanical, electrical, chemical, environmental, aerospace, materials, fire, and industrial engineering) and 55 subdomains, and is crossed with seven intent types (i.e., definition, explanation, calculation, comparison, design/synthesis, troubleshooting, and code-related) and three difficulty tiers (undergraduate, graduate, and professional), yielding 57,750 records with field/subdomain/type/difficulty metadata and solution formatting. We examined ERI via seven LLMs and report a statistically significant three-tier performance structure, with frontier models (GPT-5, Claude Sonnet 4, DeepSeek V3.1) achieving mean scores above 4.30 on a five-point scale, while mid-tier and smaller models exhibited progressively higher failure rates and steeper performance degradation on graduate-level questions. To address circularity concerns inherent in LLM benchmarks, we developed a convergent validation protocol that leverages cross-provider independe

We introduce an automatically generated benchmark for predicting hidden text in technical papers. A paper supplies visible context $X$ and a hidden continuation $Y$; the evaluated model writes an auxiliary forecast string $Z$, and a separate scorer assigns next-token probability to $Y$ both with and without conditioning on $Z$. This gives a label-free test of whether $Z$ transmits information about the continuation, compared against controls where $Z$ is recent context rather than a forecast. Our main testbed is equation-suffix prediction: the predictor sees context and the first part of a displayed equation, then forecasts the rest. The task mixes surface-level arXiv/TeX text modeling with reasoning-sensitive inference; the suffix is one of many roughly equivalent continuations, so the benchmark is read statistically rather than item-by-item. On 1363 equation continuations from 138 recent physics and mathematics papers, forecasts from GPT-5.5, Opus 4.7, and GPT-5.4 nano all improve clipped likelihood over the context control under both Qwen3-8B and Kimi K2.6 scorers, distinguishing model families and reasoning-effort settings without human labels. To emulate shortcuts where $Z$ fu

Large language models now achieve high final-answer accuracy on mathematical reasoning benchmarks, but accuracy alone does not capture reasoning flexibility. We introduce a strategy-level evaluation framework instantiated on 80 AMC 10/12 and AIME problems with 217 AoPS-derived reference strategy families. Model outputs are annotated for strategy identity, validity, and correctness using dual-AI coding with human adjudication. Across four frontier models, we find a pronounced decoupling between answer accuracy and strategy diversity. Under a single-solution prompt, all models achieve high accuracy (95%-100%), but under a multiple-strategy prompt they recover substantially fewer strategies than the human reference set. Gemini, DeepSeek, GPT, and Claude generate 184, 152, 151, and 110 distinct valid strategies, respectively, with the largest gaps in Geometry and Number Theory. The models collectively produce 50 benchmark-novel valid strategies, indicating both incomplete coverage of human strategies and some capacity for alternative reasoning. A repeated-run robustness check on 20 problems shows diminishing gains in discovered strategies, with the strongest model recovering only 39 of

Large multimodal models (LMMs) are increasingly integrated into autonomous driving systems for user interaction. However, their limitations in fine-grained spatial reasoning pose challenges for system interpretability and user trust. We introduce Logic-RAG, a novel Retrieval-Augmented Generation (RAG) framework that improves LMMs' spatial understanding in driving scenarios. Logic-RAG constructs a dynamic knowledge base (KB) about object-object relationships in first-order logic (FOL) using a perception module, a query-to-logic embedder, and a logical inference engine. We evaluated Logic-RAG on visual-spatial queries using both synthetic and real-world driving videos. When using popular LMMs (GPT-4V, Claude 3.5) as proxies for an autonomous driving system, these models achieved only 55% accuracy on synthetic driving scenes and under 75% on real-world driving scenes. Augmenting them with Logic-RAG increased their accuracies to over 80% and 90%, respectively. An ablation study showed that even without logical inference, the fact-based context constructed by Logic-RAG alone improved accuracy by 15%. Logic-RAG is extensible: it allows seamless replacement of individual components with i

Code comments are important in software development because they directly influence software maintainability and overall quality. Bad practices of code comments lead to code comment smells, negatively impacting software maintenance. Recent research has been conducted on classifying inline code comment smells, yet automatically detecting these still remains a challenge. We aim to automatically detect and classify inline code comment smells through machine learning (ML) models and a large language model (LLM) to determine how accurately each smell type can be detected. We enhanced a previously labeled dataset, where comments are labeled according to a determined taxonomy, by augmenting it with additional code segments and their associated comments. GPT 4, a large language model, was used to classify code comment smells on both the original and augmented datasets to evaluate its performance. In parallel, we trained and tested seven different machine learning algorithms on the augmented dataset to compare their classification performance against GPT 4. The performance of models, particularly Random Forest, which achieved an overall accuracy of 69 percent, along with Gradient Boosting a

Improving Large Language Model (LLM) agents for sequential decision-making tasks typically requires extensive task-specific knowledge engineering--custom prompts, curated examples, and specialized observation/action spaces. We investigate a different approach where agents automatically improve by learning from their own successful experiences without human intervention. Our method constructs and refines a database of self-generated trajectories that serve as in-context examples for future tasks. Even naive accumulation of successful trajectories yields substantial performance gains across three diverse benchmarks: ALFWorld (73% to 89%), Wordcraft (55% to 64%), and InterCode-SQL (75% to 79%). These improvements exceed those achieved by upgrading from gpt-4o-mini to gpt-4o and match the performance of allowing multiple attempts per task. We further enhance this approach with two innovations: database-level curation using population-based training to propagate high-performing example collections, and exemplar-level curation that selectively retains trajectories based on their empirical utility as in-context examples. With these enhancements, our method achieves 93% success on ALFWorld

Cricket is the second most popular sport worldwide, with billions of fans seeking advanced statistical insights unavailable through standard web searches. Although LLMs have advanced significantly in Text-to-SQL tasks, their capability to handle domain-specific nuances and multilingual requirements in sports analytics remains under-explored. We present CricBench, a benchmark suite evaluating the intrinsic SQL generation abilities of LLMs on cricket data across four formats: Test, ODI, T20I, and IPL. We curate a Gold-Standard dataset of 2,654 evaluation instances across four languages (English, Hindi, Punjabi, and Telugu). We evaluate seven models, GPT-5 Mini, Claude Sonnet 4, DeepSeek R1 and V3, Qwen 235B, Llama 3.1, and Gemma 2, using schema-only prompting. No single model dominates across all formats: GPT-5 Mini leads on Test cricket (12.4% DMA), Qwen 235B leads on IPL (28.7%) and T20I (17.5%), and all models score 0% on hard ODI queries. All models show a stark disconnect between syntactic validity (>98% execution accuracy) and semantic correctness (<29% DMA), with a domain gap of 37-55 percentage points versus BIRD. To our knowledge, CricBench is the first Text-to-SQL ben

The ubiquity and widespread use of digital and online technologies have transformed mental health support, with online mental health communities (OMHCs) providing safe spaces for peer support. More recently, generative AI and large language models (LLMs) have introduced new possibilities for scalable, around-the-clock mental health assistance that could potentially augment and supplement the capabilities of OMHCs. Although genAI shows promise in delivering immediate and personalized responses, its effectiveness in replicating the nuanced, experience-based support of human peers remains an open question. In this study, we harnessed 24,114 posts and 138,758 online community (OC) responses from 55 OMHCs on Reddit. We prompted several state-of-the-art LLMs (GPT-4-Turbo, Llama-3, and Mistral-7B) with these posts, and compared their responses to human-written (OC) responses based on a variety of linguistic measures across psycholinguistics and lexico-semantics. Our findings revealed that AI responses are more verbose, readable, and analytically structured, but lack linguistic diversity and personal narratives inherent in human--human interactions. Through a qualitative examination, we fo

Fine-tuning large language models on narrow data with harmful content produces broadly misaligned behavior on unrelated prompts, a phenomenon known as emergent misalignment. We propose that emergent misalignment involves persona-model collapse: deterioration of the model's internal capacity to simulate, differentiate, and maintain consistent characters. We test this hypothesis behaviorally using two metrics: moral susceptibility (S) and moral robustness (R), computed from the across- and within-persona variability of models' Moral Foundations Questionnaire responses under persona role-play. These metrics formalize the model's ability to differentiate characters (S) and its consistency when simulating a given one (R). We evaluate four frontier models (DeepSeek-V3.1, GPT-4.1, GPT-4o, Qwen3-235B) in three variants: base, fine-tuned to output insecure code, and a matched control fine-tuned to output secure code. Across the four models, insecure fine-tuning produces an average $55\%$ increase in S, pushing all four insecure variants beyond the band observed across 13 frontier models benchmarked in prior work -- with GPT-4o reaching more than twice the band's upper end -- signaling dysre

OCR systems, ranging from classical engines to specialised OCR vision-language models (OCR-VLMs) and frontier multimodal LLMs, report strong results on English and Chinese document benchmarks, yet their behaviour on Indic scripts is largely uncharacterised. We benchmark ten systems on Devanagari (Hindi): classical EasyOCR; open VLMs (Qwen2.5-VL-3B, Qwen3-VL-8B, olmOCR-7B); specialised OCR-VLMs (DeepSeek-OCR, Unlimited-OCR); and frontier closed models (Gemini 2.5 Flash, Claude Opus 4.7, GPT-5.5, Mistral OCR), across four synthetic degradation conditions and 300 real printed scans. We report four findings. First, on clean rendered text all ten cluster within chrF++ 91 to 98, so synthetic text does not separate them. Second, under degradation the specialised OCR-VLMs are the most fragile: DeepSeek-OCR suffers rare but catastrophic repetition failures (outputs up to 71 the reference length) that wreck its corpus mean even though its median is the best of any system, which is why we report median and catastrophic-rate instead of the mean. Third, on real scans nine of the ten systems collapse (EasyOCR falls from chrF++ 93.6 to 58.3) and the field spreads across a 76-point range, so synth