Search AI Products and News

Liquid AI opensources the next-generation edge AI model LFM2, available in three versions with 350M to 1.2B parameters. The model features an innovative architecture, achieving twice the inference speed and three times the training efficiency on edge devices, supporting 32K long context processing. LFM2 performs exceptionally well in tasks such as instruction following, outperforming models of similar scale, making it particularly suitable for privacy-sensitive scenarios. Fully open-sourced through Hugging Face, this marks the first time a U.S. company has surpassed Chinese open-source models in the field of efficient small models. Liquid AI

A significant breakthrough in the field of artificial intelligence. AIbase learned from social media that ByteDance recently announced the open-sourcing of its new multimodal generation model, Liquid. This model, utilizing an innovative unified encoding method and a single large language model (LLM) architecture, seamlessly integrates visual understanding and generation tasks. This release not only showcases ByteDance's technological ambition in multimodal AI but also provides a powerful open-source tool for global developers. Below is AIbase's in-depth analysis of the Liquid model, exploring its technological innovations and core features.

Google DeepMind recently released a comprehensive strategic document outlining its approach to developing safe Artificial General Intelligence (AGI). AGI is defined as a system capable of matching or exceeding human capabilities in most cognitive tasks. DeepMind anticipates that current machine learning approaches, particularly neural networks, will remain the primary path to achieving AGI. The report notes that future AGI systems could potentially surpass human performance and exhibit significant autonomy in planning and decision-making. This technology will have implications for healthcare...

Tianyancha App shows that Beijing Silicon-based Liquid Technology Co., Ltd. recently underwent a major business change. In this change, Beijing Zhipu AI Technology Co., Ltd. withdrew from the shareholder ranks, while four new investors were added: Meituan's Tianjin SanKuai Technology Co., Ltd., Beijing Xinglian Dingsen Equity Investment Fund Partnership (Limited Partnership), Nanjing LvYong Jinhao Equity Investment Management Partnership (Limited Partnership), and SME Development Fund PwC (Hangzhou) Venture Capital Partnership (Limited Partnership). After this round of financing, the registered capital of Silicon-based Liquid increased from approximately RMB 13.758 million...

The AI revolution in image recognition is rapidly evolving, moving beyond simple cat-dog classification. Current trends focus on much finer distinctions, such as identifying the year and model of a specific car or even subtle differences like the thickness of a bird's eyebrow. However, while neural networks excel at classification, explaining their reasoning ('Why did you classify this as X?') often proves challenging. Traditional Class Activation Maps (CAM)

Researchers from the Tübingen University's ELLIS Unit, the University of Maryland, and Lawrence Livermore National Laboratory have developed Huginn, a novel language model with a recursive architecture that significantly enhances reasoning capabilities. Unlike traditional models, Huginn doesn't require specialized 'reasoning chain' training; it can autonomously reason within the neural network's 'latent space' before outputting results. The research team developed Huginn, a novel language model with a recursive architecture that significantly enhances reasoning abilities. Unlike...

Google Research recently announced the innovative 'Titans' series model architecture, achieving a groundbreaking 2 million token context length through bionic design, with plans to open-source related technologies in the future. The core innovation of this architecture is the introduction of a Deep Neural Long-Term Memory Module, inspired by the human memory system. Titans cleverly combines the rapid response capability of short-term memory with the persistence characteristics of long-term memory, while utilizing an attention mechanism to handle immediate context, forming an efficient information processing system.

A recent study suggests that artificial intelligence (AI) has the potential to surpass human intelligence in the future. The research delves into the complexities of AI and the human brain and proposes a new theoretical framework, indicating that by applying AI technology at the cellular level of neuroscience, AI could infinitely approach the functionality of the human brain, potentially even surpassing human intelligence levels. The human brain is widely regarded as one of the most complex systems in the universe, while artificial intelligence is seen as one of the most important technologies in human history. A core question looms: Will artificial intelligence ultimately surpass human intelligence?

Meta has recently launched a new neural watermark system called VideoSeal, designed to help identify AI-generated videos, even if the videos have been edited, it can effectively trace their origin. This watermark is invisible to viewers but can be verified through detection methods to confirm the video's authenticity. Meta has released the entire system as open source software under the MIT license, including the VideoSeal watermark model, research papers, and training and inference code.

At the recent annual re:Invent conference held in Las Vegas, Amazon's cloud computing service AWS announced a series of new technologies, highlighting the energy efficiency improvements of its data centers, achieving efficiency levels over four times greater than traditional on-premises computing. AWS head Prasad Kalyanaraman stated that these innovations not only enhance energy efficiency but also flexibly support emerging workloads. Image source note: Image generated by AI, image licensed by service provider Midj.

AI painting is a revolutionary image generation technology based on deep learning algorithms, particularly Generative Adversarial Networks (GANs) and diffusion models. This approach analyzes vast amounts of image data to learn and simulate human painting techniques, thereby creating entirely new visual works. AI painting not only captures and reproduces the complex details of the real world accurately but also blends different artistic styles, showcasing astonishing creativity and imagination. The core of this technology lies in transforming abstract textual descriptions into concrete visual expressions.