On March 10th, Zhiyuan Robotics officially launched its first general-purpose embodied base large model – Genie Operator-1 (GO-1). This announcement has garnered significant attention, particularly regarding its potential in home service robots and the new hope it offers for future household management. According to Zhiyuan Robotics' official introduction, the GO-1 large model, trained on a vast amount of human video data, demonstrates excellent performance in executing various household tasks such as delivering cups, preparing meals, and greeting guests. In terms of technical performance,

The internal testing of the AI agent Manus has sparked significant excitement, with users praising its groundbreaking capabilities and demanding access. Its innovative features are revolutionizing the traditional virtual assistant experience.

The ByteDance Doubao large model team announced today the successful development of a new sparse model architecture called UltraMem. This architecture effectively addresses the high memory access issues during the inference of MoE (Mixture of Experts) models, improving inference speed by 2 to 6 times compared to MoE, and reducing inference costs by up to 83%. This groundbreaking advancement opens a new path for efficient inference of large models. The UltraMem architecture successfully resolves the memory bottleneck during inference of MoE architectures while maintaining model performance. Experimental results show that the parameters and activation conditions are the same.

Recently, Meta has announced a groundbreaking research achievement by developing a new type of memory layer technology that can significantly enhance the factual accuracy of large language models (LLMs) and achieve unprecedented expansion in parameter scale. This technology not only challenges traditional neural network expansion methods but also provides new directions for future AI architecture design. The core of this research lies in utilizing a trainable key-value lookup mechanism to add extra parameters to the model without increasing computational load (FLOPs). This method's core thinking is...

Against the backdrop of the global generative AI wave in 2022, YuanShi Intelligence (RWKV) completed tens of millions of RMB in angel round financing in December 2023, with investment from Tianji Capital. Following this financing, the company’s valuation doubled, and the funds will be used for team expansion, new architecture development, and product commercialization. The emergence of RWKV presents a strong challenge to the traditional Transformer architecture. With the advancement of large language models (LLMs), although the parameter scale of models is becoming increasingly large, issues like hallucination and accuracy persist.

Yuanshi Intelligence Co., Ltd. in Shenzhen recently announced that it has completed tens of millions of RMB in angel round financing on December 25, 2024, with Tianji Capital as the investor. This financing will accelerate Yuanshi Intelligence's efforts in ecological development, strengthening technological capabilities, and launching AI applications targeting end consumers. Founded in June 2023, Yuanshi Intelligence is a high-tech enterprise focusing on the research and development of large model architectures and AI applications. With its innovative technology and clear development strategy, the company has continuously attracted the favor of the capital markets. Prior to this financing, Yuanshi Intelligence had

Since Microsoft first launched the C o p i l o t Plus laptop based on Arm architecture in June, there has been anticipation about whether the features of C o p i l o t Plus could be applied to desktop computers. Six months later, the upcoming mini PC from Microsoft will bring AI capabilities to Windows 11, such as smart reminders, click-to-do lists, and AI image generation and editing. These mini PCs are expected to help Microsoft compete with Apple's latest Mac.

The highly anticipated Deepseek V3 is finally open source! This brand-new AI model has achieved significant breakthroughs in multilingual programming capabilities, even outperforming competitors like Claude 3.5 Sonnet V2 in the aider multilingual programming assessment, attracting widespread attention in the industry. It is reported that Deepseek V3 has made a qualitative leap in performance compared to previous versions. Deepseek V2.5 had a success rate of only 17% in the aider assessment, while V3 has surged to 4.

Kimi Technology Co., Ltd. and Tsinghua University's MADSys laboratory have jointly released an open-source project called Mooncake, aimed at collaboratively building a large model inference architecture centered around KVCache. In June 2024, both parties released the design plan for the Mooncake inference system based on Kimi, which significantly enhances inference throughput through PD separation and a storage-computation architecture, garnering widespread attention in the industry.

In an era of rapid development in artificial intelligence, the intelligence level of large models continues to improve, but the challenges of inference system efficiency are becoming increasingly apparent. Addressing high inference loads, reducing inference costs, and shortening response times has become an important issue faced by the industry. Kimi has collaborated with the MADSys laboratory at Tsinghua University to launch the KVCache-based Mooncake inference system design plan, which will be officially released in June 2024. The Mooncake inference system enhances efficiency through its innovative approach.

Today, OPPO announced that it has successfully applied the Expert Mixture of Experts (MoE) architecture to mobile devices for the first time globally. This groundbreaking advancement significantly enhances AI processing efficiency, opening new pathways for the further development of terminal AI technologies and laying the foundation for the deep integration of AI and mobile hardware. As AI technology rapidly advances, an increasing number of AI tasks need to be completed on terminal devices. However, large AI models often require powerful computing capabilities, which poses a huge challenge for mobile devices with limited hardware resources. This breakthrough aims to overcome this limitation.

In today's rapidly advancing field of artificial intelligence, a multi-modal large language model named ORYX is quietly transforming our understanding of AI's ability to perceive the visual world. This AI system, developed collaboratively by researchers from Tsinghua University, Tencent, and Nanyang Technological University, is regarded as the 'Transformers' of visual processing. ORYX, short for Oryx Multi-Modal Large Language Models, is an AI model specifically designed for processing images, videos, and 3D scene time-space understanding.

Recently, NVIDIA's CEO Jensen Huang revealed in an interview that OpenAI's latest o1 large language model (LLM) will be based on NVIDIA's Blackwell architecture, which has achieved an astounding improvement in inference capability. As the market gradually emerges from the hype of generative AI, developing LLM models capable of generating 'inference-focused' responses has become increasingly important, marking a significant step toward achieving general artificial intelligence.

At the Baidu Cloud Smart Conference held on September 25, 2024, Baidu Intelligent Cloud announced a series of significant upgrades for enterprise AI applications, aimed at helping businesses utilize large models more efficiently and economically. These upgrades include version 4.0 of the Baijie AI Heterogeneous Computing Platform and version 3.0 of the Qianfan Large Model Platform, as well as a comprehensive update of AI native application products such as the Code Assistant, Intelligent Customer Service, and Digital Humans.

AWPortrait-FL is an image generation model focused on portrait photography, trained on FLUX.1-dev. It significantly enhances image composition and detail representation, particularly skin texture and realism, by using a dataset of high-aesthetic quality fashion photography. Generated images are limited to non-commercial use in accordance with a non-commercial license agreement. Key features of the model include high-quality image generation, improved skin detail, flexible prompting capabilities, GPU acceleration, a non-commercial use license, various resolution options, and excellent performance.