# nlp Published articles for nlp. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps DevFeed: [Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps]() Original publisher: [Read original article]() Author: Leonie Monigatti; ben burtenshaw; Sergio Paniego Published: 2026-09-03T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [AI Chat]() Tags: [artificial-intelligence](), [benchmark](), [community](), [evaluation](), [fine-tuning](), [gpu](), [grpo](), [guide](), [json](), [llama-cpp](), [llm](), [model](), [nlp](), [open-source](), [python](), [rl](), [training](), [trl]() ### AI overview A tutorial on fine-tuning a 350M language model with GRPO to improve structured-output and JSON Schema compliance, then evaluating it on the IFStruct benchmark. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers DevFeed: [Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers]() Original publisher: [Read original article]() Author: Tom Aarsen Published: 2026-08-26T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [sentence-transformers](), [AI search](), [AI, ML & Data Engineering]() Tags: [community](), [datasets](), [embedding](), [embeddings](), [evaluation](), [guide](), [nlp](), [open-source](), [retrieval](), [sentence-transformers](), [training]() ### AI overview A tutorial on finetuning multi-vector embedding models with Sentence Transformers. It explains late-interaction token-level retrieval, training components, and domain-specific retrieval improvements. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers DevFeed: [Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers]() Original publisher: [Read original article]() Author: Tom Aarsen; Antoine Chaffin; Raphael Sourty Published: 2026-08-18T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [sentence-transformers](), [Embeddings](), [Open Source]() Tags: [community](), [embedding](), [embeddings](), [guide](), [multimodal](), [nlp](), [open-source](), [retrieval](), [search](), [sentence-transformers](), [transformers](), [vector]() ### AI overview This article explains multi-vector, or late-interaction, embedding models with Sentence Transformers. It covers token-level representations, MaxSim scoring, retrieval over text and page images, integration with search systems, and index-size tradeoffs. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## DCD: доменно-ориентированная архитектура для построения RAG-систем DevFeed: [DCD: доменно-ориентированная архитектура для построения RAG-систем]() Original publisher: [Read original article]() Author: redmadrobot (red\_mad\_robot) Published: 2026-06-18T09:42:21Z Content type: article Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Retrieval Augmented Generation (RAG)]() Tags: [collection](), [domain](), [genai](), [llm](), [nlp](), [rag](), [tag-1cd610c0e518](), [tag-68e701e78517](), [tag-7bc388df28ed](), [tag-d89cae10e887]() ### AI overview The article introduces DCD (Domain-Collection-Document), a hierarchical architecture for organizing knowledge and processing queries in RAG systems. It describes domain-based search restriction, multi-stage routing, structured model outputs, smart chunking, hybrid search, and validation mechanisms intended to improve retrieval and generation quality in heterogeneous document collections. ### Source excerpt Привет! Это Роботы. Недавно мы выпустили статью на arXiv, где представили архитектурный подход DCD (Domain-Collection-Document) для структурирования пространства знаний и обработки запросов в RAG-системах. Мы провели подробные эксперименты, оценили работу подхода на собственном бенчмарке и показали, как он влияет на качество поиска и генерации в сравнении с другими подобными методами. А теперь хотим здесь рассказать о ключевых идеях, лежащих в основе DCD Design. Читать далее ## Beyond LoRA: Can you beat the most popular fine-tuning technique? DevFeed: [Beyond LoRA: Can you beat the most popular fine-tuning technique?]() Original publisher: [Read original article]() Author: Benjamin Bossan; Sayak Paul; Marian Tietz; Kashif Rasul Published: 2026-06-18T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [AI, ML & Data Engineering](), [SDKs]() Tags: [artificial-intelligence](), [community](), [diffusers](), [fine-tuning](), [guide](), [lora](), [nlp](), [peft](), [quantization](), [stable-diffusion](), [training](), [transformers]() ### AI overview The article examines parameter-efficient fine-tuning (PEFT) for open models, focusing on whether LoRA is always the best technique. It explains how PEFT reduces memory requirements, can enable fine-tuning of quantized models, and offers a unified API for multiple techniques. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Introducing the Ettin Reranker Family DevFeed: [Introducing the Ettin Reranker Family]() Original publisher: [Read original article]() Author: Tom Aarsen Published: 2026-05-19T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [sentence-transformers](), [Transformer](), [long-context](), [Agent Skill]() Tags: [ai-coding](), [community](), [long-context](), [nlp](), [open-source](), [ranking](), [research](), [retrieval](), [sentence-transformers]() ### AI overview The article introduces six Sentence Transformers CrossEncoder rerankers built on Ettin ModernBERT encoders. It explains their distillation-based training, retrieval-then-rerank usage, quality and cost trade-offs, and support for up to 8K tokens of context. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Эволюция данных: генетический алгоритм в задаче классификации текстов DevFeed: [Эволюция данных: генетический алгоритм в задаче классификации текстов]() Original publisher: [Read original article]() Author: evgeneration (red\_mad\_robot) Published: 2026-04-23T19:05:42Z Content type: article Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Natural language processing](), [Large Language Model]() Tags: [ai](), [evolution](), [llm](), [ml](), [nlp](), [prompt](), [tag-1cd610c0e518](), [tag-1f12060c781a](), [tag-2d1b71a8786f](), [tag-7bc388df28ed](), [tag-9bf5e01ce62e](), [tag-9d8cf70dc46c]() ### AI overview The article describes using a genetic algorithm to evolve prompts that generate training data for text-classification models. The approach was motivated by poor results from synthetic datasets and manual prompt iteration, with prompt fitness evaluated by classifier performance on a fixed test set. ### Source excerpt Привет! Меня зовут Женя Андриевская. Я -- NLP-инженер в лаборатории R&D red_mad_robot. Сегодня расскажу, как мы призвали на помощь эволюцию, чтобы улучшить качество данных в задачах классификации текстов. Да-да, ту самую эволюцию, с мутациями и скрещиваниями -- только в мире данных, промптов и LLM. Читать далее ## QIMMA قِمّة ⛰: A Quality-First Arabic LLM Leaderboard DevFeed: [QIMMA قِمّة ⛰: A Quality-First Arabic LLM Leaderboard]() Original publisher: [Read original article]() Author: Leen AlQadi; Ahmed Alzubaidi; Mohammed Alyafeai; Maitha Alhammadi; Shaikha Alsuwaidi; Omar saif alkaabi; Basma Boussaha; Hakim Hacid Published: 2026-04-21T10:09:58Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [Ground truth / benchmark quality](), [LLM evaluation / benchmarking](), [Benchmark](), [LLMs](), [Natural language processing](), [datasets](), [Inference](), [Open Source](), [Code]() Tags: [benchmark](), [benchmarks](), [blog](), [code](), [data](), [evaluation](), [hugging-face](), [inference](), [leaderboard](), [llm](), [nlp](), [open-source]() ### AI overview QIMMA is a quality-first Arabic LLM leaderboard that validates benchmark data before evaluating models. It addresses translation issues, annotation errors, encoding problems, cultural bias, reproducibility gaps, and fragmented task coverage. The platform combines native Arabic content, systematic validation, code evaluation, and public per-sample inference outputs across 109 subsets from 14 benchmarks and more than 52,000 samples. ### Source excerpt A Blog post by Technology Innovation Institute on Hugging Face ## Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers DevFeed: [Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers]() Original publisher: [Read original article]() Author: Tom Aarsen Published: 2026-04-16T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [multimodal](), [sentence-transformers](), [datasets](), [Machine learning](), [Artificial Intelligence](), [Open Source]() Tags: [community](), [data](), [datasets](), [embedding](), [evaluation](), [models](), [multimodal](), [nlp](), [open-source](), [optimization](), [performance](), [retrieval](), [sentence-transformers](), [train](), [training](), [transformers](), [vision]() ### AI overview A practical guide to fine-tuning multimodal Sentence Transformer embedding and reranker models for visual document retrieval. It explains the training pipeline and shows how domain-specific fine-tuning improved retrieval performance from 0.888 to 0.947 in the example evaluation. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Fine-tuning a DistilBERT classifier with numerical and text inputs DevFeed: [Fine-tuning a DistilBERT classifier with numerical and text inputs]() Original publisher: [Read original article]() Author: Paloma Jol Published: 2026-04-10T10:04:33Z Content type: tutorial Language: en Sources: [FreeAgent]() Topics: [Fine-tuning](), [Transformer](), [foundation-models](), [Training AI Models](), [datasets](), [Kaggle](), [scikit-learn]() Tags: [ai](), [aws](), [data](), [data-ml](), [data-science](), [dataset](), [encoding](), [fine-tuning](), [hugging-face](), [kaggle](), [llms](), [machine-learning](), [model](), [models](), [nlp](), [performance](), [text-classification](), [training-data](), [validation]() ### AI overview This tutorial explains how to fine-tune a DistilBERT text classifier while incorporating a numerical feature into the same network. It uses the Kaggle wine reviews dataset, combining review descriptions with price and preprocessing the data for training, validation, and testing. ### Source excerpt Text classification is often done through fine-tuning of a pretrained foundation model with domain-specific data. In FreeAgent we use transformer based models to automatically classify incoming bank transactions. Specifically we use a DistilBERT model that is fine-tuned on hundreds of millions of bank transactions with customer-labelled accounting categories. The model inputs are currently text-based, built from a combination of bank transaction descriptions and amounts. In this post we describe an approach to fine-tuning the DistilBERT model and training the classifier including the numerical amount feature as a single network. ## Multimodal Embedding & Reranker Models with Sentence Transformers DevFeed: [Multimodal Embedding & Reranker Models with Sentence Transformers]() Original publisher: [Read original article]() Author: Tom Aarsen Published: 2026-04-09T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [sentence-transformers](), [Embeddings](), [multimodal](), [Retrieval Augmented Generation (RAG)](), [vlm](), [GPU](), [Cloud]() Tags: [cloud](), [community](), [embedding](), [gpu](), [multimodal](), [nlp](), [open-source](), [rag](), [retrieval](), [search](), [sentence-transformers](), [transformers](), [vision]() ### AI overview This article explains how multimodal embedding and reranker models in Sentence Transformers map text, images, audio, and video into shared spaces or score cross-modal pairs. It covers visual document retrieval, cross-modal search, multimodal RAG, hardware requirements, model loading, and similarity computation. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Natural Language Processing (NLP) Series DevFeed: [Natural Language Processing (NLP) Series]() Original publisher: [Read original article]() Author: baeldung Published: 2026-04-01T15:52:18Z Content type: tutorial Language: en Sources: [Baeldung - CS]() Topics: [Natural language processing]() Tags: [artificial-intelligence](), [artificial-intelligence-series-natural-language-processing](), [guide](), [natural-language-processing](), [nlp](), [series]() ### AI overview A guide series covering the core areas of natural language processing. ### Source excerpt This guide walks through the core areas of NLP. The post Natural Language Processing (NLP) Series first appeared on Baeldung on Computer Science. Related Stories What Is Aspect-Based Sentiment Analysis? What Are Language Models in NLP? Sorting Algorithms Series ## Как маскировать персональные данные на изображениях: наш эксперимент с OCR и NER DevFeed: [Как маскировать персональные данные на изображениях: наш эксперимент с OCR и NER]() Original publisher: [Read original article]() Author: andrivasg (red\_mad\_robot) Published: 2026-03-17T15:55:37Z Content type: tutorial Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Natural language processing](), [pii]() Tags: [ai](), [computer-vision](), [llm](), [ner](), [nlp](), [ocr](), [pii](), [red-mad-robot](), [rnd](), [tag-601fbc7112a4](), [tag-7bc388df28ed](), [tag-9bf5e01ce62e](), [tag-b92bf5906bbd](), [tag-ef0b1bf200df]() ### AI overview The article describes red_mad_robot's experiment using OCR combined with a NER model to detect and selectively mask personally identifiable information in images without training specialized visual detectors. On a dataset of 40 annotated images, the pipeline masked 90% of personal data while falsely masking 14% of text polygons; performance declined on difficult real-world photographs. ### Source excerpt Всем привет! Меня зовут Андрей Иванов, я NLP-исследователь в R&D red_mad_robot. Мы разрабатываем систему Guardrails для защиты персональных данных (PII) и фильтрации небезопасного контента. В этой статье расскажу, как мы решали задачу точечного маскирования PII на картинках без обучения специальных визуальных детекторов. Разберём связку оптического распознавания символов (OCR) с NER-моделью, покажем метрики на реальных данных, раскроем ограничения подхода и наши решения для их преодоления. Читать далее ## Highlights of Booking.com's publication in 2025 DevFeed: [Highlights of Booking.com's publication in 2025]() Original publisher: [Read original article]() Author: Yang Yang Published: 2026-01-20T09:20:13Z Content type: article Language: en Sources: [Booking.com Data Science]() Topics: [Machine learning](), [Artificial Intelligence](), [Natural language processing](), [Language models](), [Fine-tuning](), [recommendation systems](), [Optimization](), [Compression](), [Inference]() Tags: [2025](), [ai](), [booking](), [compression](), [data-science](), [fine-tuning](), [inference](), [language-models](), [latency](), [neurips](), [nlp](), [optimization](), [publication]() ### AI overview Booking.com highlights its 2025 machine learning publications, including papers accepted at major conferences and research on applying Medusa speculative decoding and knowledge distillation to travel-related language model tasks. ### Source excerpt At Booking.com, our mission is to make experiencing the world easier for everyone. We are committed to investing in cutting-edge technology that removes the barriers to travel, enabling seamless connections between millions of travelers and unforgettable experiences, diverse transportation options, and exceptional accommodations. The intersection of academic rigor and industry application is where true transformation happens. In 2025, our ML community bridged this gap more effectively than ever, contributing vital new insights to the global scientific community. With 8 out of 13 papers accepted at premier conferences -- including NeurIPS, SIGIR, KDD, and ACL -- our colleagues have demonstrated world-class expertise in AI, NLP, recommendation systems, uplift modeling, etc. These aren't just theoretical wins; they are the engines of innovation that allow us to push technological boundaries, ensuring our platform remains the most sophisticated and intuitive guide in the ever-evolving travel industry. Below, we highlight some of the key achievements and insights from these groundbreaking works. Speed Without Sacrifice: Fine-Tuning Language Models with Medusa and Knowledge Distillation in Travel Applications By Daniel Zagyva, Emmanouil Stergiadis, Laurens Van Der Maas, Aleksandra Dokic, Eran Fainman, Ilya Gusev, Moran Beladev Best paper award of 2025 ACL Industry Track https://aclanthology.org/2025.acl-industry.48/ In high-stakes industrial NLP applications, balancing generation quality with speed and efficiency presents significant challenges. We address them by investigating two complementary optimization approaches: Medusa for speculative decoding and knowledge distillation (KD) for model compression. We demonstrate the practical application of these techniques in real-world travel domain tasks, including trip planning, smart filters, and generating accommodation descriptions. We introduce modifications to the Medusa implementation, starting with base pre-trained models ## MCP Tool Registry: автоматизированное создание систем RAG DevFeed: [MCP Tool Registry: автоматизированное создание систем RAG]() Original publisher: [Read original article]() Author: maksimov\_m (red\_mad\_robot) Published: 2025-12-30T16:07:23Z Content type: tutorial Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Model Context Protocol](), [Retrieval Augmented Generation (RAG)](), [Large Language Model](), [Natural language processing](), [Artificial Intelligence](), [Tool]() Tags: [ai](), [llm](), [mcp](), [model-context-protocol](), [nlp](), [rag](), [tag-41e9f0e0da78](), [tag-68e701e78517](), [tag-db2a5a2010aa](), [tag-ebcf3e175cd3](), [tool](), [workflows]() ### AI overview This Russian-language developer article presents MCP Tool Registry, a central registry for managing MCP servers used to build Retrieval Augmented Generation systems. It explains how RAG and Model Context Protocol address LLM limitations around context and access to external tools and data, and describes registry-based orchestration of multi-step workflows across multiple servers. ### Source excerpt Всем привет! Меня зовут Максим Максимов, я -- NLP-инженер в AI R&D лаборатории в red_mad_robot. Мы работаем с прикладными AI-системами, проводим эксперименты и проверяем архитектурные подходы в реальных сценариях. Современные AI-архитектуры всё чаще выносят прикладную логику за пределы LLM -- в инструменты, внешние источники данных и правила их оркестрации. Модель выступает вычислительным ядром, а ценность формируется на уровне системы вокруг неё. Из этого подхода в R&D-лаборатории вырос MCP Tool Registry -- центральный реестр для управления системой серверов, необходимых для построения RAG. В этой статье я подробно расскажу, из каких частей состоит MCP Tool Registry, как он работает и покажу его применение на практическом примере. Читать далее ## Контроль против гибкости: два подхода к созданию AI-агентов DevFeed: [Контроль против гибкости: два подхода к созданию AI-агентов]() Original publisher: [Read original article]() Author: Mentalitet (red\_mad\_robot) Published: 2025-11-14T17:15:42Z Content type: article Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Artificial Intelligence](), [Large Language Model](), [AI Development](), [Natural language processing](), [Langgraph](), [SDKs](), [OpenAI](), [cursor]() Tags: [agent](), [agents](), [ai](), [cursor](), [development](), [development-kit](), [langgraph](), [llm](), [llm-05dacf5e57d9](), [llm-12323b29df45](), [nlp](), [openai](), [red-mad-robot](), [sdk](), [software-development](), [tag-1605473766c5](), [tag-61cd5a476b1d](), [tag-68e701e78517](), [tag-93edcb27480f](), [tag-b0a411324cb6](), [tag-dd24610a6838]() ### AI overview The article compares code-first and graph-first approaches to building multi-agent systems. It describes how LLMs coordinate tools and other agents, including ReAct, handoff, agent-as-tool, and group-chat patterns, drawing on the author's development experience and documentation from OpenAI and LangGraph. ### Source excerpt Привет! Меня зовут Женя Орлов, я руковожу NLP-разработкой в red_mad_robot. Мы экспериментируем с мультиагентными системами и изучаем, как LLM ведут себя в разных архитектурах. В процессе накопилось много наблюдений и рабочих инсайтов -- хочу поделиться тем, что помогает нам при проектировании AI-агентов. В последнее время внимание инженеров и исследователей привлекают агентные архитектуры -- системы, где LLM координирует работу внутренних инструментов и других моделей. Разберём два основных подхода к построению агентных систем. Читать далее ## Sentence Transformers is joining Hugging Face! DevFeed: [Sentence Transformers is joining Hugging Face!]() Original publisher: [Read original article]() Author: Tom Aarsen Published: 2025-10-22T00:00:00Z Content type: news Language: en Sources: [Hugging Face - Blog]() Topics: [sentence-transformers](), [hugging face](), [Embeddings](), [Natural language processing](), [Open Source]() Tags: [announcement](), [collaboration](), [community](), [contributors](), [embeddings](), [hugging-face](), [natural-language-processing](), [nlp](), [open-source](), [sentence-transformers]() ### AI overview Sentence Transformers, also known as SentenceBERT or SBERT, is joining Hugging Face. The popular open-source library generates embeddings that capture semantic meaning and supports natural language processing tasks such as semantic search, similarity analysis, clustering, and paraphrase mining. It will remain community-driven under the Apache 2.0 license, with contributions welcomed from researchers, developers, and enthusiasts. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## AI Agents and Multi-Agent Systems: A Practical Case for Analyzing and Fixing Code Vulnerabilities DevFeed: [AI Agents and Multi-Agent Systems: A Practical Case for Analyzing and Fixing Code Vulnerabilities]() Original publisher: [Read original article]() Author: redmadrobot (red\_mad\_robot) Published: 2025-10-16T15:18:12Z Content type: tutorial Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Artificial Intelligence](), [Natural language processing](), [Large Language Model]() Tags: [agents](), [ai](), [ai-agents](), [graph](), [llm](), [nlp](), [sast](), [systems](), [tag-1605473766c5](), [tag-4d5248fd6bfe](), [workflow]() ### AI overview The article explains the defining characteristics and autonomy levels of AI agents, focusing on workflow agents and multi-agent systems. It presents a practical case involving a multi-agent environment for automatically analyzing and fixing code vulnerabilities. ### Source excerpt Привет! Меня зовут Егор Козлов, я работаю NLP-инженером в red_mad_robot. Мы активно внедряем в бизнес AI-агентов -- автономных и полуавтономных программных сущностей, которые самостоятельно выполняют задачи и принимают решения в интересах бизнеса. В статье расскажу о принципах работы AI-агентов -- с особым вниманием к workflow-агентам и мультиагентным системам (MAS). И поделюсь практическим кейсом внедрения мультиагентной среды для автоматического анализа и исправления уязвимостей в коде. Читать далее ## Introducing RTEB: A New Standard for Retrieval Evaluation DevFeed: [Introducing RTEB: A New Standard for Retrieval Evaluation]() Original publisher: [Read original article]() Author: Frank Liu; Kenneth Enevoldsen; Solomatin Roman; Isaac Chung; Tom Aarsen; Fődi, Zoltán Published: 2025-10-01T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [Benchmark](), [LLM evaluation / benchmarking](), [Retrieval Augmented Generation (RAG)](), [generalization in machine learning](), [recommendation systems]() Tags: [artificial-intelligence](), [benchmark](), [benchmarks](), [community](), [datasets](), [developers](), [embedding](), [evaluation](), [models](), [nlp](), [open](), [performance](), [rag](), [recommendation-systems](), [research](), [retrieval](), [search](), [zero-shot]() ### AI overview Hugging Face introduces the beta Retrieval Embedding Benchmark (RTEB), designed to evaluate the retrieval accuracy and generalization of embedding models in real-world applications. It combines open and private datasets to provide a fairer, more transparent, application-focused evaluation standard. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Accelerating Qwen3-8B Agent on Intel® Core™ Ultra with Depth-Pruned Draft Models DevFeed: [Accelerating Qwen3-8B Agent on Intel® Core™ Ultra with Depth-Pruned Draft Models]() Original publisher: [Read original article]() Author: Igor Margulis; Ofir Zafrir; Shira Guskin; Guy Boudoukh; Pedro Cuenca Published: 2025-09-29T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [qwen](), [AI Agent](), [Inference](), [intel](), [smolagents](), [GPU](), [Latency]() Tags: [ai-agent](), [gpu](), [guide](), [hub](), [inference](), [intel](), [latency](), [llm](), [models](), [nlp](), [qwen](), [smolagents]() ### AI overview The article explains how to accelerate the Qwen3-8B agent on an Intel Lunar Lake integrated GPU using OpenVINO.GenAI. Speculative decoding with Qwen3-0.6B as a draft model achieves about a 1.3x speedup, while pruning the draft model increases the speedup to about 1.4x. It also demonstrates running a fast, local AI agent with smolagents. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Detecting hallucinations in RAG systems with circuit tracing and attribution graphs DevFeed: [Detecting hallucinations in RAG systems with circuit tracing and attribution graphs]() Original publisher: [Read original article]() Author: IrinaKoshkina (red\_mad\_robot) Published: 2025-09-26T18:30:09Z Content type: tutorial Language: ru Sources: [Redmadrobot EN](), [Redmadrobot RU]() Topics: [Large Language Model](), [Retrieval Augmented Generation (RAG)](), [Natural language processing](), [anthropic]() Tags: [anthropic](), [circuit](), [llm](), [nlp](), [rag](), [red-mad-robot](), [tag-0444141ce356](), [tag-3f4d1152a2b4](), [tag-73f5f2c76797](), [tag-b6914c0b0244](), [tag-d27a0708d400](), [tracing](), [transcoder]() ### AI overview The article explains an approach to detecting hallucinations in retrieval-augmented generation systems by examining model computation graphs with Anthropic's framework, cross-layer transcoders, and attribution graphs. It presents an implementation for analyzing Qwen2.5-7B and reports 85% detection accuracy on a test dataset for a simple implementation. ### Source excerpt Всем привет! Меня зовут Ирина, я NLP-инженер в red_mad_robot, занимаюсь научными исследованиями интерпретируемости LLM и анализом механизмов внутренних вычислений моделей, чтобы применять полученные результаты на практике. Например, сегодня хочу рассказать, как мы подошли к решению задачи детекции галлюцинаций LLM в RAG системах со стороны исследования графов размышлений модели -- с помощью интересного фреймворка от Anthropic. В статье поговорим, как использовать cross-layer transcoders и атрибуционные графы для детекции галлюцинаций в RAG системах. Рассмотрим пример реализации детектора на графах для анализа модели Qwen2.5-7B и практические примеры использования. Спойлер: на простой реализации получили точность детекции 85% на тестовом датасете. Читать далее ## mmBERT: ModernBERT goes Multilingual DevFeed: [mmBERT: ModernBERT goes Multilingual]() Original publisher: [Read original article]() Author: Marc Marone; Orion Weller; William Fleshman; Eugene Yang; Dawn Lawrie; Ben Van Durme Published: 2025-09-09T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [Artificial Intelligence](), [Training AI Models](), [Machine Learning & Artificial Intelligence](), [datasets](), [Open Source Models & Datasets](), [Data Quality](), [Boilerplate]() Tags: [artificial-intelligence](), [blog-post](), [community](), [data-quality](), [datasets](), [llm](), [model](), [models](), [nlp](), [open-source](), [open-source-collab](), [performance](), [research](), [speed](), [training]() ### AI overview This blog post introduces mmBERT, a massively multilingual encoder model trained on more than 3 trillion tokens across over 1,800 languages. It describes performance and speed improvements over earlier multilingual models, the model's ModernBERT-based architecture, and a progressive strategy for adding languages and balancing training data. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Welcome EmbeddingGemma, Google's new efficient embedding model DevFeed: [Welcome EmbeddingGemma, Google's new efficient embedding model]() Original publisher: [Read original article]() Author: Tom Aarsen; Joshua; Alvaro Bartolome; Aritra Roy Gosthipaty; Pedro Cuenca; Sergio Paniego Published: 2025-09-04T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [Embeddings](), [sentence-transformers](), [Artificial Intelligence](), [Google](), [Retrieval-Augmented Generation](), [Open Source](), [transformers.js](), [Inference](), [onnx](), [dataset](), [LangChain](), [llamaindex]() Tags: [artificial-intelligence](), [benchmark](), [community](), [embedding](), [google](), [guide](), [langchain](), [llamaindex](), [nlp](), [on-device](), [onnx](), [open-source](), [rag](), [sentence-transformers](), [transformers-js]() ### AI overview Google introduces EmbeddingGemma, a compact multilingual embedding model designed for fast, efficient on-device use. The article covers its architecture, training, multilingual capabilities, benchmark performance, framework integrations, and domain fine-tuning for retrieval applications. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## 🇵🇭 FilBench - Can LLMs Understand and Generate Filipino? DevFeed: [🇵🇭 FilBench - Can LLMs Understand and Generate Filipino?]() Original publisher: [Read original article]() Author: Lj V. Miranda; Elyanah Aco; Conner Manuel; Jan Christian Blaise Cruz; Joseph Imperial; Daniel van Strien; Nathan Habib; Clémentine Fourrier Published: 2025-08-12T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [Language models](), [Human-AI evaluation]() Tags: [cebuano](), [community](), [evaluation](), [filipino](), [generate](), [generation](), [inference](), [llm](), [llms](), [nlp](), [open-source](), [philippines](), [tagalog](), [testing]() ### AI overview FilBench is an evaluation suite for assessing how LLMs understand and generate Tagalog, Filipino, and Cebuano. It evaluates cultural knowledge, classical NLP, reading comprehension, and translation-oriented generation across 12 tasks. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. [Next page]()