Natural Language Processing news and analysis on Web Pulse: 5424 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
Independent LLM-from-scratch experiments found that OpenAI's GPT-2 small (124M parameters) consistently outperformed the author's 163M-parameter models on an instruction fine-tuning task adapted from Sebastian Raschka's …
Developer Beatriz Almeida released papero, an open-source PDF parser that extracts reading order, tables, formulas, figures and bounding boxes from PDFs using geometry alone, with no ML models and CPU-only execution. The…
A developer is building PlanToday, an AI-powered event and wedding planning platform that converts natural-language requirements — such as a wedding in Delhi for 300 guests on a ₹15 lakh budget — into structured data for…
Inception released Mercury Voice, a diffusion LLM tuned for voice agents, to general availability for enterprise customers, claiming a median (p50) time to first answer token of under 320 milliseconds on production custo…
A developer built Typic, a 400M-parameter open decision model trained on a single free Kaggle T4 GPU in 46 minutes, which returns probabilities over a supplied list of allowed answers in one forward pass instead of gener…
DeepL updated its privacy policy to remove the sales(at)deepl.com contact address for data processing agreements under Art. 28 GDPR, directing users instead to privacy@deepl.com. The policy states that content uploaded t…
ServiceNow Inc. launched Flow, a conversational natural language interface for its service desk that handles employee requests with extensive automation, the company said. ServiceNow is targeting organizations that want …
A developer hardened a URL-to-Markdown extraction API after auditing request logs and finding that roughly one in six URLs returned under 60 words of valid-looking Markdown — navigation menus and footers from JavaScript-…
Laya released a multilingual, non-autoregressive "System 1" decision engine that returns typed decisions across 100+ languages in a single forward pass in 33 ms, installable via `python -m pip install laya` on Python 3.1…
A developer guide argues that teams building AI meeting-notes features should buy hosted transcription from vendors like Deepgram and AssemblyAI and build only the summarization, decision-extraction and CRM/ticketing int…
David AI's DAI-ASR-I18N benchmark, built by CEO Tomer Cohen and CTO Ben Wiley, compared 14 speech-recognition systems on unscripted conversation across 21 languages and found error rates varied by an average of 3.7 perce…
NVIDIA published a fine-tuning workflow for its Nemotron 3.5 ASR streaming 0.6B model that adapts the 40-language-locale speech recognizer to Saudi Najdi and Hijazi dialects using the NeMo framework, the SADA 2022 and FL…
GoldiMask, a fine-tuning method for discrete diffusion language models introduced in arXiv paper 2609.38385v1, selects which response tokens to reveal as context by approximately maximizing a submodular objective and the…
Split-conformal claim filtering raises the share of multi-hop RAG responses whose retained claims are fully supported to 95.80%-97.20% at a 95% target, up from 55.60%-76.03% without filtering, according to an arXiv paper…
A study posted to arXiv (2609.38261v1) introduced NinaXander, a method that connects layers of frozen language models from different architecture families through a single trained shared-latent adapter, composing the rec…
A new arXiv paper (2609.38406v1) introduces a ~9.2K-instance multi-domain benchmark for narrative text infilling and reports that model scale does not reliably predict infilling quality, with Gemma-2-2B scoring highest q…
A new arXiv paper, arXiv:2609.38374v1, proposes Doc2LoRA, a method that represents each scientific paper as a LoRA adapter generated by a Doc-to-LoRA hypernetwork, so that every point in the paper space — including mixtu…
Researchers introduced STAM, a state-transition-aware memory framework for LLM clinical agents that records state changes as new clinical entries arrive, combining semantic retrieval with typed clinical relations to keep…
Team MSU placed third in the ACM MM 2026 GenText-Forensics challenge with a decomposed chain-of-thought pipeline that pairs a document tampering detector with two Qwen3-VL-32B vision-language models, each LoRA-adapted to…
Researchers introduced MOBA-VL, a 9B-parameter vision-language model trained with event-localized multi-turn reinforcement learning on game telemetry to deliver real-time commentary for Multiplayer Online Battle Arena es…