small language model

The Universal Algorithm: Physics Walls, Collective Solutions, and the Architecture of Complexity

THE UNIVERSAL ALGORITHM Physics Walls, Collective Solutions, and the Architecture of Complexity The Universal Algorithm is the general-audience synthesis of a developing research programme at the intersection of complexity science, evolutionary theory and artificial life. The programme asks whether major evolutionary t...

What beliefs are most debunkable?

Large language models (LLMs) can persuade people to revise their beliefs, but it is unclear which beliefs are most "debunkable," and why. Across three pooled studies, participants (N = 3,341 online U.S. adults) discussed a self-reported conspiracy or other epistemically suspect belief with GPT-4, instructed to debunk i...

WITFuzz: Validity-Preserving Greybox Fuzzing for WebAssembly Interface Type Binding Generators

Modern build pipelines often rely on code generation to turn constraint-rich interface specifications into artifacts for target programming languages. In the WebAssembly component model, binding generators (bindgens) follow this pattern by translating WebAssembly Interface Types (WIT) packages into language-specific bi...

Persistence and Recoverability of Correct Information After Recognition Errors in Dysarthric Speech Recognition

Automatic speech recognition (ASR) often produces incorrect 1-best transcriptions for dysarthric speech, but a recognition error does not necessarily imply that all information supporting the correct utterance has been lost. This study investigates the extent to which correct information remains within an ASR model aft...

A Staged Anti-Phishing Browser Extension for One-Day Scam Pages

Legion is an extension for Chrome, Edge, and Brave. It looks at the page a person already has open and warns them when that page asks for a password, a card, an identity document, or a one-time code and behaves like a one-day phishing page. Such pages collect data in the first hours, while their addresses are still abs...

OpenEMS/openems: 2026.10.0

Release Highlights New implementations: Phoenix Contact PLCnext Alfen NG9xx Wallbox SAX Power Home Plus Sungrow PV Inverter & ESS Hardy Barth cPH1 Improved implementations: FENECON & GoodWe Battery Inverter KEBA P40 FENECON Commercial 100 Fronius Modbus/TCP API Controller REST API PV-Inverter SellToGridLimit Controller...

Empowering Lightweight Language Models for Security Code Review via Context-Aware Distillation

Security code review, which specifically examines software from a security perspective, is indispensable for preempting vulnerabilities and bolstering software reliability. However, existing automated approaches face a dilemma. They either rely on large language models with prohibitive deployment costs, or employ light...

Tokenizer Impact on Multilingual to Portuguese Transfer in ModernBERT-Based Models

PTIT-CourseQA: A Vietnamese University Course-Material QA Benchmark, Annotations and KG-RAG Evaluation Outputs

PTIT-CourseQA is a Vietnamese question-answering benchmark over 11 university textbooks of the Posts and Telecommunications Institute of Technology (PTIT) in four fields (information security, information technology, electronics, marketing). It contains 1,000 test and 200 development questions of four types (single-hop...

Named Entity Recognition Across Datasets and Domains: Resources and Models for Galician

Automatic named entity recognition (NER) is essential for many natural language processing applications, particularly in low-resource languages and those with corpora restricted to a single domain, where the lack of diverse data may hinder cross-domain generalisation. In this context, we attempt to contribute to the...

How Technological Innovation Restructures Cultural Space

What this is. The film-level data behind the article How Technological Innovation Restructures Cultural Space: Digital Effects, Narrative Convergence, and the Concentration of Economic and Symbolic Capital in American Film, 1980–2024 (Poetics). It covers 7,868 U.S. films with production-technology records: VFX technolo...

Benchmark Validity in Financial Language-Model Evaluation

Benchmark results are increasingly used to compare language models for financial decision-support tasks, but such comparisons can confound domain learning with compliance with task-specific output contracts. This study examines that measurement problem using FinVector-Market-4B, a LoRA adaptation of Qwen3.5-4B for stru...