Natural Language Processing news and analysis on Web Pulse: 2449 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
ViToS, a dual-stream reinforcement learning framework, prunes unnecessary visual tokens in medical imaging to streamline vision-language models. It reduces inference load and boosts performance, achieving up to 108.27% r…
Researchers have developed a CVE-TTP Knowledge Graph linking software vulnerabilities to attacker behaviors using the MITRE ATT&CK framework. Transformer-based models like CySecBERT achieved high F1-scores for identifyin…
LeapXpert raised $180 million in a growth round led by Riverwood Capital on June 30, 2026, to convert governed business messaging data into AI-ready intelligence for regulated enterprises. The funding brings total equity…
Transcribe.cpp, a C/C++ speech-to-text inference library, has been released supporting 16 model families and 60+ variants via GGUF models on the ggml runtime. It offers Metal, Vulkan, and CUDA backends for GPU accelerati…
A developer building an LLM-based invoice data extraction system found that naive prompting led to frequent hallucinations and only 60-70% accuracy. By switching to a validated generation approach using Pydantic schemas …
Bol.ai launches a tool that extracts structured data from Bills of Lading, commercial invoices, and packing lists using AI, enabling freight forwarders and customs brokers to automate data entry. The platform supports PD…
Researchers introduced the Relative Surprisal Index (RSI), an information-theoretic metric that balances token probability and entropy to improve reinforcement learning for large language models. The RSI Selection method…
Researchers developed CSTrader, a multi-agent system using large language models to trade Counter-Strike 2 weapon skins, outperforming a down-trending market index by 15.62% and achieving up to 7.58% cumulative returns d…
Researchers released HistoriQA-ThirdRepublic, a French-language dataset of 1,782 multi-hop questions designed to test AI's ability to reason across parliamentary debates and newspapers from the French Third Republic. Cre…
A developer built @ttsc/graph, an MCP tool that provides coding agents with a TypeScript compiler-built code graph instead of raw source files. In benchmarks, it reduces token usage by roughly 10x on structural questions…
A developer built iamuu, a free online image and PDF processing platform using Vue 3, FastAPI, Celery, and MinIO, offering over 100 image format conversions, 40 PDF tools, AI background removal, OCR, and digital signatur…
Siplinx AI launched a meeting notetaker that runs local LLM and STT models on-device, offering real-time transcription, summaries, and action items without cloud processing. The tool supports Zoom, Google Meet, and other…
A new research framework called CLExEval reveals that large language models like GPT-4o-mini and HuatuoGPT-o1 exhibit significant flaws in clinical reasoning, including verbosity bias, hidden knowledge paradox, and a mis…
Researchers introduced Latent Ordinal Prototype Alignment (LOPA) and Semantic-Anchored Layer Routing (SALR), a new framework for spoken language assessment that matches the performance of billion-parameter models using a…
A new study introduces the Guided-Retry strategy to reduce AI hallucinations in task-oriented dialogues without retraining models. Tested on models like DeepSeek-R1 and Llama-3, the method cut hallucination rates by up t…
A new AI model using Multi-View Gated Graph Attention Networks achieved 90% accuracy in detecting Alzheimer's disease through spontaneous speech analysis on the ADReSSo dataset. The model integrates semantic, dependency,…
Researchers introduced Triospect, a new AI-text detection framework that improves detection accuracy by 22.3% AUROC and 13% TPR01 on the Humanize-16K dataset, and shows resilience against sophisticated attacks. The frame…
Researchers have proposed a new evaluation protocol for speech-to-speech AI that focuses on conversational prosody and rhythm, using over 4000 hours of dyadic English conversations to create matched reference regimes. Th…
A new natural language-to-SQL agent using a semantic-layer-mediated approach achieved 94.15% execution accuracy on the Spider2-snow benchmark, ranking third on the official leaderboard. The system decouples semantic inte…
Researchers introduced LoFa, a new benchmark to evaluate large language models' resistance to logical fallacies, revealing inconsistent robustness across models. The study highlights the risk of deploying AI systems vuln…