Author: kissdev

In a rapidly evolving tech landscape, artificial intelligence (AI)-powered eyewear has surfaced as the latest trend, captivating Chinese tech firms eager to meld generative AI with wearable devices. This surge of interest follows the release of Meta smart glasses by Ray-Ban, which set the precedent for the integration of AI in everyday accessories. Superhexa, a Xiaomi-backed start-up, has made its foray into the AI eyewear market with the launch of Jiehuan, its AI audio glasses, this month. Priced competitively at 699 yuan (approximately US$98), these glasses offer functionality akin to their more expensive overseas counterparts. With built-in speakers and microphones,…

TEL AVIV, Israel, Aug. 22, 2024 /PRNewswire/ — AI21, a trailblazer in the development of foundational models and AI systems for enterprises, has unveiled two groundbreaking additions to its Jamba model family: Jamba 1.5 Mini and Jamba 1.5 Large. These models promise unparalleled quality and low latency, furnished with the largest context windows currently available. With their revolutionary architecture, the new Jamba models eclipse others in their class, outperforming even behemoths like Llama 8B and 70B. Building on the success of their predecessors, these enhanced models signify a substantial advancement in long-context language models, offering unmatched speed, efficiency, and performance…

In the rapidly evolving landscape of artificial intelligence, grounding techniques such as retrieval-augmented generation (RAG) have gained substantial traction to prevent AI models from producing erroneous or “hallucinated” outputs. However, even the most well-implemented RAG models have their shortcomings. Giants in technology like Google and Microsoft are now pioneering new grounding methodologies to enhance the accuracy and timeliness of AI systems. Imagine providing an AI model with a map to navigate the world. What happens when that map becomes outdated or lacks critical details? This is the conundrum that hyperscalers and leading AI developers are striving to resolve. As artificial…

OpenAI recently rolled out a significant update to its artificial intelligence (AI) toolkit, introducing the capability to fine-tune GPT-4o, its most advanced language model to date. This eagerly awaited feature allows developers to adapt the model to meet specific business needs, setting the stage for a new era of tailored AI applications across various industries. The Importance of Fine-Tuning Fine-tuning is a critical process where a pretrained AI model is customized for specific tasks or specialized areas. According to OpenAI, “Developers can now fine-tune GPT-4o with custom datasets to achieve higher performance at a lower cost for their unique use…

By Ross Pomeroy, RealClearWire In recent years, large language models (LLMs) have become integral to daily life. Whether they’re powering chatbots, digital assistants, or guiding us through internet searches, these sophisticated artificial intelligence (AI) systems are becoming increasingly ubiquitous. LLMs, which ingest vast amounts of text data to learn and form associations, can produce a variety of written content and engage in surprisingly competent conversations with users. Given their expanding role and influence, it’s crucial these AI systems remain politically neutral, especially when tackling complex political issues. However, a recent study published in PLoS ONE indicates otherwise. David Rozado, an…

In the ever-evolving field of Artificial Intelligence, Large Language Models (LLMs) have been the center of significant research, with continuous efforts being made to enhance their performance across an array of tasks. A foremost challenge in this endeavor is understanding how pre-training data influences the models’ overall capabilities. Although the value of diverse data sources and computational resources has been acknowledged, a pivotal question remains unresolved: what intrinsic properties of the data most effectively bolster general performance? Interestingly, code data has emerged as a potent component in pre-training mixtures, even for models that aren’t primarily utilized for code generation tasks.…

Nvidia and Mistral AI have unveiled a groundbreaking compact language model that boasts “state-of-the-art” accuracy in a remarkably efficient package. This new marvel, the Mistral-NemMo-Minitron 8B, is a streamlined iteration of the NeMo 12B, having been reduced from 12 billion to 8 billion parameters. In a blog post, Bryan Catanzaro, Vice President of Deep Learning Research at Nvidia, explained that this downsizing was achieved through two sophisticated AI optimization methods: pruning and distillation. Pruning involves trimming the neural network by removing the weights that minimally affect accuracy. Following this, the team employed a distillation process, retraining the pruned model on…

In an era where data reigns supreme, Snowflake is at the forefront of a revolution, leveraging the power of artificial intelligence to transform raw data into actionable insights and sophisticated applications. This innovative approach was recently showcased in an episode of NVIDIA’s AI Podcast. Here, host Noah Kravitz delved into the intricacies of Snowflake’s AI Data Cloud platform with Baris Gultekin, Snowflake’s head of AI, offering listeners a compelling glimpse into a future sculpted by advanced AI technologies. Redefining Data Management Snowflake’s AI Data Cloud platform promises to redefine how enterprises handle and exploit their data. By decoupling data storage…

In the ever-evolving arena of artificial intelligence, Chinese powerhouses Baidu and SenseTime, accompanied by the innovative start-up Zhipu AI, stand as the foremost providers of business-centric large language model (LLM) services in China. This assertion comes from a pioneering report by market research firm IDC, which underscores the relentless pursuit of generative AI integration within the tech landscape. Baidu AI Cloud emerges as the leader in China’s industry-focused LLM market, commanding a significant 19.9 percent market share and yielding an impressive revenue of 350 million yuan (approximately US$49 million) in 2023. SenseTime follows closely, securing the second position with a…

The growing discourse around AI technologies often zeroes in on their expansive potential and transformative impacts. However, Darren Oberst, CEO of Ai Bloks and the AI framework platform LLMWare, argues that smaller, localized versions of AI language models could be essential in tackling burgeoning concerns related to data privacy and the costs associated with these technologies. Speaking at the Forward Festival in Madison, Oberst shed light on how these compact models could be the unsung heroes of the AI world. Hosted by the MadAI group, a community of AI professionals in the Madison area, his insights offered a refreshing take…