Natural Language Processing news and analysis on Web Pulse: 5440 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
HeronLoom, an open-source pipeline from developer korries, combines iterative clustering, LLM-based analysis, graph construction, and analytical search to turn social-media posts or documents into a structured, explorabl…
Shrijith Venkatramana, a developer building the AI code review tool LiveReview, argues that RAG pipelines can return results in milliseconds yet still surface the wrong documents, and that embedding evaluation should be …
A developer built the Multilingual Code and Language Dataset, an open-source collection of nearly 150 GB of uncompressed text and source code covering Central Asian languages including Kyrgyz, Kazakh, Uzbek, and Tajik, n…
A maintainer of the Whoosh pure-Python full-text search library published a runnable walkthrough of phrase and proximity queries, showing how quoted searches like "machine learning" match only adjacent, in-order terms wh…
A developer built jevgrep, a semantic code search tool that skips the vector index entirely by combining a ripgrep keyword sweep with a single batched call to TypeSafe's Jev decision model, which returns a probability di…
SafeSeal, a patent-pending LLM watermarking technology available for licensing at TRL 3, embeds identifiable marks in large language model outputs while achieving a BERTScore of 0.981, an entity similarity score of 0.962…
A developer has published a technical explainer on tokenization, detailing how large language models convert text into integer token sequences rather than processing words directly. The writeup covers subword tokenizatio…
Classifier.dev launched a free, keyless zero-shot text classification API that it says outperforms the Jev model it runs on, scoring 90.0% on AG News versus 87.5% for Jev alone and 62.7% on emotion versus 61.8%, measured…
A practical guide explains how large language models work for product managers without requiring deep ML expertise, walking through the pipeline from tokenization and embeddings through the Transformer's self-attention a…
Linkup released SPARSEUP, an open-source learned sparse embedding model with a 149M-parameter ModernBERT backbone, published under Apache 2.0 on Hugging Face. Linkup reports 56.4 average nDCG@10 on BEIR-13, which it call…
Famulor's September 17, 2026 update lets teams prepare and preview fixed announcements — greetings, consent messages, and end-call farewells — as saved audio generated in the background from saved assistant and voice set…
Tesla's 2026 Summer Update, software version 2026.26.6.5, gave the Grok voice assistant direct control of vehicle hardware, and a community test reported by Drive Tesla Canada found a driver posting as LifeMiddie execute…
The LectuLibre team built a token-based chunking pipeline for translating full-length books with large language models, using Python, FastAPI, and Anthropic's Claude alongside DeepSeek for simpler passages. The system sp…
Researchers Mohamed Suliman and colleagues published an arXiv paper on 30 October 2022, revised 9 October 2023, showing that federated learning does not protect user privacy in Google's GBoard next word prediction model.…
A developer explains embeddings as the mechanism that lets search match text with no shared words, such as linking "how do I reset my password" to a page titled "account recovery." The walkthrough describes how models li…
Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two live-dialogue AI models that can hold real-time conversations while handling visual context and executing background tasks. The standard model …
SpaceXAI released Grok Voice Transcribe 2.0 on September 18th, claiming the speech recognition model is twice as accurate as version 1.0 at the same price of $0.10 per batch hour and $0.20 for streaming, with Atlassian s…
Researchers Tianyu Fu and colleagues posted Cache-to-Cache (C2C) to arXiv on 3 Oct 2025, revised 2 Mar 2026, a method that lets large language models communicate directly through fused KV-cache representations instead of…
Rijul, a developer building the AI code review tool LiveReview, outlined why vector search alone is insufficient for retrieval-augmented generation systems and argued for hybrid retrieval. The writeup describes combining…
A developer published an explainer arguing that large language models are fundamentally next-token probability functions rather than systems that know facts or reason before answering. The piece traces hallucination, cha…