THE INTERNET IS THINKING
Earlier sources: Language models
Language models learn patterns in text and other representations to predict or generate sequences. Follow model releases, training methods, reasoning, evaluation, and deployment research; a benchmark result is specific to its test conditions.
SYMTRI organizes observations from public sources into a living map. Connections reflect shared subject matter; they do not establish causation or endorse a source.
Open in the map ↗461 classified signals in the fourteen-day window. These counts include signals awaiting readable evidence.
Earlier sources
Up to ten sources with readable evidence, ordered by their feed timestamp. These archive sources can be older than the live fourteen-day window. This list is a sample, not a measure of total activity. Dates distinguish paper publication, repository creation, model/dataset activity, Hacker News posts, and KEV catalog additions. “First archived” records when SYMTRI first stored the signal.
- arXiv · Published · First archived
CyberPersistBench: Evaluating LLM-Based Cyber Attackers on Installation and Persistence ↗
While LLM-based attackers exhibit growing proficiency in vulnerability exploitation, most existing cybersecurity benchmarks suffer from single-stage truncation, prematurely terminating evaluation upon initial access. In practice, initial footholds are exceptionally fragile across operational disruptions such as service restarts and host reboots. Whether LLM-based attackers can establish and maintain durable footholds beyond initial compromise remains a central blind spot in cybersecurity evaluation. To bridge this gap, we introduce CyberPersistBench, the first benchmark dedicated to post-compr…
- arXiv · Published · First archived
Know the Normal, Track the Attack: Context-Grounded and Stateful LLM Investigation over System Provenance ↗
Provenance-based intrusion detection systems (PIDSs) identify suspicious activity in audit streams, but their outputs remain difficult to turn into coherent attack narratives. Direct LLM analyses of local anomalous subgraphs lack deployment-specific normal-behavior knowledge and validated attack state across evidence fragments. This can cause unsupported attack interpretations of routine activities and incorrect attribution of temporally dispersed evidence to attack stages. We present ANCHOR, an investigation-oriented provenance system that combines evidence curation with context-grounded LLM…
- GitHub · Repository created · First archived
B0yko/local-enough ↗
Measure whether a back-office AI task can run on your own hardware: accuracy vs gold, p50/p95 latency, $/1k tasks, amortised hardware cost and break-even volume, plus an OpenAI-compatible router that sends each task to the cheapest model meeting your quality bar.
- OpenAlex · Published · First archived
Traceable Cybersecurity Conformity Pre-Assessment of Electric Vehicle Charging Equipment Using Cross-Standard Gap Analysis and a Local LLM ↗
Electric vehicle charging equipment now links vehicles, payment and roaming services, cloud platforms, and the grid, but its cybersecurity requirements are fragmented across standards. This study combines a cross-standard gap analysis with a traceable conformity pre-assessment framework for a local large language model; the framework is a design-validation proof of concept, not a production-ready conformity-assessment system. Using EN 18031 as the reference axis, IEC 62443-4-2, ETSI EN 303 645, and NIST IR 8473 requirements were mapped on a five-level correspondence scale by a single researche…
- OpenAlex · Published · First archived
Rethinking Attention: Entropy-Guided Iterative Refinement for Language Models ↗
Standard Multi-Head Attention (MHA) computes contextual representations in a single forward pass, providing no mechanism to detect or correct diffuse or suboptimal attention distributions. Existing efforts to improve attention have primarily targeted computational efficiency or sequence length, leaving attention quality itself largely unaddressed. We propose IMHA (Entropy-Guided Iterative Multi-Head Attention), a within-layer attention formulation that progressively refines token representations over T iterations using shared projection weights. At each step, the Shannon entropy of each token’…
- OpenAlex · Published · First archived
A Scalable Method for Validated Data Extraction from Electronic Health Records with Large Language Models ↗
Purpose/Background: Health care organizations increasingly require structured, patient-level clinical variables for treatment decisions, operational workflows, quality measurement, and clinical trial screening. Relevant information is often fragmented across heterogeneous electronic health record (EHR) systems, unstructured formats, and scanned documents. Large language models (LLMs) offer an opportunity to enhance medical record extraction, particularly in oncology where point-of-care structured coding rarely captures a patient’s full longitudinal clinical course. Methods: Two complementary L…
- OpenAlex · Published · First archived
LLM-supported self-diagnosis and self-treatment of familial Mediterranean fever: a case report ↗
Large language models (LLMs) are increasingly being used by patients to seek explanations for persistent or unexplained symptoms. However, their impact on real-world diagnostic pathways and patient behavior remains poorly understood. We report the case of a 34-year-old male of Egyptian origin with recurrent episodes of fever, severe migratory abdominal pain, arthralgia, myalgia, and intermittent erythematous skin lesions since early childhood. Despite almost 3 decades of extensive investigations, no unifying diagnosis had been established. During an iterative dialogue with the LLM DeepSeek in…
- OpenAlex · Published · First archived
Fine-Grained Analysis of Bilingual Transportation Discourse on Reddit Using Large Language Models: A Quebec City Case Study ↗
Social media platforms generate large volumes of unsolicited textual content that reflect users’ perceptions of urban transportation. This study proposes a fine-grained annotation framework that classifies social media text into three constructs: attitude, intention, and reported behavior. Each statement is linked to a specific transportation mode and sentiment polarity. Two large language models (LLMs), GPT-5.5 and DeepSeek-Flash, were evaluated using few-shot in-context learning on 200 manually annotated English–French Reddit discussions from Quebec City, Canada. GPT-5.5 was selected based o…
- OpenAlex · Published · First archived
Evaluating intent-driven color generation: a framework for assessing LLM-assisted cartographic design ↗
Large language models (LLMs) offer new possibilities for adaptive cartographic design, yet how they interpret user intent remains unclear. This study evaluates how stated analytical goals influence LLM-generated color schemes for thematic maps, assessing three perceptual dimensions: whether colors form a clear sequence matching the data hierarchy (order), whether categories are visually distinct from one another (differentiation), and whether related values share similar or thematically cohesive colors (association). Using a synthetic dataset of 20 lakes classified by pollution level (Low, Med…
- OpenAlex · Published · First archived
Frequency control of a two-area RES-integrated power system incorporating a smart transformer using a fractional-order controller ↗
The growing demand for electricity, together with high penetration of renewable energy sources (RES), poses serious operational issues, such as power generation fluctuation, reduced system inertia, and unstable control performance. This article investigates a two-area automatic generation control (AGC) system comprising a thermal system (TS) along with renewable energy sources (RES) that include a solar thermal power plant (STPP) and wind turbine system (WTS), utilizing smart transformers (STs). The article proposed a new fractional order proportional integral minus tilt integral derivative (F…
Live window recorded . This is the ingestion observation time, not the publication or discovery date of each source.
How to assess evidence about Language models
Check each original source for its methods, release status, limitations, and supporting data. A repository announcement, a preprint, and a published study provide different kinds of evidence. SYMTRI’s subject classification helps discovery; it does not independently validate the source’s claims.
See the collection and counting methodology ↗ · Read the weekly synthesis ↗
Related threads
Explore related regions
These navigation suggestions are part of the map’s editorial structure, rather than measured evidence of a connection.