Latest news with #languageModel
Yahoo
4 days ago
- Automotive
- Yahoo
Mitsubishi Electric Develops Edge-device Language Model for Domain-specific Manufacturing
Leverages data augmentation to optimize language-model responses for user applications TOKYO, June 18, 2025--(BUSINESS WIRE)--Mitsubishi Electric Corporation (TOKYO: 6503) announced today that it has developed a language model tailored for manufacturing processes operating on edge devices. The Maisart®-branded AI technology has been pre-trained with data from Mitsubishi Electric's internal operations, enabling it to support a wide range of applications in specific manufacturing domains. In addition, the model leverages a uniquely developed data-augmentation technique to generate responses optimized for user-specific applications. The widespread adoption of generative AI is accelerating the use of large language models (LLMs). However, the significant computational and energy costs associated with LLMs are a growing concern. Additionally, there is increasing demand for generative AI solutions that can operate in on-premises environments due to data privacy and confidential information management requirements. In response, Mitsubishi Electric has developed a domain-specific language model by training a publicly available Japanese base model with the company's proprietary data from its own business domains, including factory automation (FA). Using training data generated through the company's original augmentation techniques enabled effective, task-specific fine-tuning. The resulting model is compact enough to run on limited hardware resources, making it suitable for environments with constrained computing capabilities such as edge devices, as well as for on-premises operations such as call centers that handle sensitive customer information. For the full text, please visit: View source version on Contacts Customer Inquiries Information Technology R&D CenterMitsubishi Electric Media Inquiries Takeyoshi KomatsuPublic Relations DivisionMitsubishi Electric CorporationTel: + Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
Bloomberg
5 days ago
- Business
- Bloomberg
Tiny Startup Brings Homegrown AI to Mongolia, Avoiding China-US Dominance
Welcome to Tech In Depth, our daily newsletter with reporting and analysis about the business of tech from Bloomberg's journalists around the world. Today, Saritha Rai looks at an AI startup in tiny Mongolia and its surprising development of a homegrown large language model. Washington Post hack: The Washington Post is investigating a cyberattack on the email accounts of some journalists, prompting the newspaper to beef up its online security.
Bloomberg
5 days ago
- Business
- Bloomberg
China's MiniMax Says Its New AI Reasoning Model Beats DeepSeek
Chinese AI upstart MiniMax released a new large language model, joining a slew of domestic peers inspired to surpass DeepSeek in the field of reasoning AI. The Shanghai-based company touted the efficiency of its new MiniMax-M1 model in handling complicated productivity tasks, claiming it outdoes all closed-source competitors from China in a statement. In several benchmarks presented by MiniMax, M1 also scored higher than DeepSeek's latest R1-0528 model.
Zawya
11-06-2025
- Business
- Zawya
TII announces Falcon-H1 model availability as NVIDIA NIM to deliver sovereign AI at scale
Paris, France - Abu Dhabi, UAE: Abu Dhabi's Technology Innovation Institute (TII), a leading global research center and the developer behind the globally ranked Falcon open-source AI models and privacy-preserving technologies, today announced that Falcon-H1, its next-generation, hybrid-architecture large language model, will be available as an NVIDIA NIM microservice. The announcement, timed with NVIDIA's GTC Paris showcase, positions Falcon-H1 for seamless enterprise deployment across cloud, on-premise, or hybrid environments. Developers can soon access and scale Falcon-H1 with production-grade performance, without the engineering overhead typically required to adapt open-source models for real-world application. Dr. Najwa Aaraj, CEO of TII, commented: 'Falcon-H1's availability on NVIDIA NIM reflects our ongoing leadership in shaping the future of open, sovereign, and cross-domain deployment ready AI. It demonstrates that breakthrough innovation from our region is not only competitive on the global stage - it's setting new benchmarks for scalable, secure, and enterprise-ready AI.' At the heart of Falcon-H1 is a novel hybrid Transformer–Mamba architecture, combining the efficiency of state space models (SSMs) with the expressiveness of Transformer networks. Designed in-house by TII researchers, the architecture supports context windows of up to 256k tokens, an order-of-magnitude leap in long-context reasoning, while preserving high-speed inference and reduced memory demands. Multilingual by design, Falcon-H1 delivers robust performance ahead of models in its category, across both high- and low-resource languages, making it suited for global-scale applications. Supported soon for deployment via the universal LLM NIM microservice, Falcon-H1 becomes a plug-and-play asset for enterprises building agentic systems, retrieval-augmented generation (RAG) workflows, or domain-specific assistants. Whether running with NVIDIA TensorRT-LLM, vLLM, or SGLang, NIM abstracts away the underlying inference stack, enabling developers to deploy Falcon-H1 in minutes using standard tools such as Docker and Hugging Face, with automated hardware optimization and enterprise-grade SLAs. 'Falcon-H1's availability on NVIDIA NIM bridges the gap between cutting-edge model design and real-world operability. It combines our hybrid architecture with the performance and reliability of NVIDIA microservices. Developers can integrate Falcon-H1 optimized for long-context reasoning, multilingual versatility, and real-world applications. What once required weeks of infrastructure tuning becomes achievable in minutes at scale, with multilingual depth, and production resilience', said Dr. Hakim Hacid, Chief AI Researcher at TII. The release also mark Falcon-H1's integration with NVIDIA NeMo microservices and NVIDIA AI Blueprints, giving developers access to full lifecycle tooling, from data curation and guardrailing to continuous evaluation and post-deployment tuning. Crucially, this makes Falcon-H1 viable in regulated, latency-sensitive and sovereign AI contexts, with full-stack NVIDIA support. With over 55 million downloads to date, the Falcon series has become one of the most widely adopted open-source models from the Middle East region. Beyond its scale, Falcon-H1 smaller variants routinely outperform larger peers on reasoning and mathematical tasks, while the 34B model now leads several industry benchmarks. TII's strategic alignment with NVIDIA's validated deployment framework reflects that open-source models are production-ready assets. Falcon-H1's availability on NIM cements its place among them as a sovereign, scalable, and secure alternative to closed-weight incumbents. For technical documentation, deployment guides and model access, Falcon-H1 is available via the Falcon LLM portal, and will be supported by an upcoming release of the universal LLM NIM microservice container at About the Technology Innovation Institute: The Technology Innovation Institute (TII) is the dedicated applied research pillar of Abu Dhabi's Advanced Technology Research Council (ATRC). TII is a pioneering global research and development center that focuses on applied research and new-age technology capabilities. The Institute has 10 dedicated research centers in advanced materials, autonomous robotics, cryptography, AI and digital science, directed energy, quantum, secure systems, propulsion and space, biotechnology, and renewable and sustainable energy. By working with exceptional talent, universities, research institutions, and industry partners from all over the world, TII connects an intellectual community and contributes to building an R&D ecosystem that reinforces the status of Abu Dhabi and the UAE as a global hub for innovation. For more information, visit Media contacts: TII@
Reuters
09-06-2025
- Business
- Reuters
Rednote joins wave of Chinese firms releasing open-source AI models
BEIJING, June 9 (Reuters) - China's Rednote, one of the country's most popular social media platforms, has released an open-source large language model, joining a wave of Chinese tech firms making their artificial intelligence models freely available. The approach contrasts with many U.S. tech giants like OpenAI and Google (GOOGL.O), opens new tab, which have kept their most advanced models proprietary, though some American firms including Meta (META.O), opens new tab have also released open-source models. Open sourcing allows Chinese companies to demonstrate their technological capabilities, build developer communities and spread influence globally at a time when the U.S. has sought to stymie China's tech progress with export restrictions on advanced semiconductors. Rednote's model, called is available for download on developer platform Hugging Face. A company technical paper describing it was uploaded on Friday. In coding tasks, the model performs comparably to Alibaba's Qwen 2.5 series, though it trails more advanced models such as DeepSeek-V3, the technical paper said. RedNote, also known by its Chinese name Xiaohongshu, is an Instagram-like platform where users share photos, videos, text posts and live streams. The platform gained international attention earlier this year when some U.S. users flocked to the app amid concerns over a potential TikTok ban. The company has invested in large language model development since 2023, not long after OpenAI's release of ChatGPT in late 2022. It has accelerated its AI efforts in recent months, launching Diandian, an AI-powered search application that helps users find content on Xiaohongshu's main platform. Other companies that are pursuing an open-source approach include Alibaba ( opens new tab which launched Qwen 3, an upgraded version of its model in April. Earlier this year, startup DeepSeek released its low-cost R1 model as open-source software, shaking up the global AI industry due to its competitive performance despite being developed at a fraction of the cost of Western rivals.
