


Youked builds a kilo-calorie inference cluster for Zhipu AI to help global users enjoy large-model smart life
Back to one night in 2021, a mother fell into a creative bottleneck and was unable to continue her novel; the father was busy writing code, envisioning developing a small game after work, but was troubled by code debugging problems; and Their children, facing a Mathematical Olympiad problem on the desk, frowned and thought hard.
Today in 2024, the emergence of large AI models changes all this.
With the help of "Zhipu Qingyan", my mother's novel creation has taken on a new lease of life. She only needs to input her creativity and ideas into the large model to generate Natural and vivid storylines and dialogues; Dad uses large models for code programming and debugging. By analyzing the code logic, he greatly reduces the tedious development process and reduces the workload by more than half; the large models have also become a learning tool for children. The assistant can not only perform intelligent homework corrections, but also provide detailed problem-solving ideas, greatly improving learning efficiency.
Large model computing power allows global users to enjoy intelligent life
Zhipu AI is committed to building the world's leading recognition platform Zhizhi Intelligent Large Model, the performance of its new generation base large model GLM-4 has been greatly improved, approaching GPT-4, demonstrating the industry's leading multi-modal large language model capabilities. Through the powerful combination of the large model of Intelligent Spectrum and the computing power of Youked, GLM-4 runs stably and efficiently on the cloud, and has large-scale real-time reasoning capabilities, successfully achieving a balance between cost-effectiveness and service quality. This innovation enables the smart spectrum model to deeply understand user needs and respond quickly, allowing users around the world to enjoy the convenience and efficiency of intelligent life in advance.
As early as 2022, Youked has begun to provide powerful underlying computing power support for Zhipu AI. Ucarte's low-cost, high-value-added Ulanqab Intelligent Computing Center provides customized high-power cabinets and abundant GPU computing power, which can help quickly build large-scale intelligent models, expand the scale of training and inference clusters, and improve models. R&D efficiency, supporting the rapid launch of large model applications and external services. At present, the total computing power management scale of Ukede Intelligent Computing Center exceeds 3000P.
Ukede Ulanqab Intelligent Computing Center
Ukerde helps Zhipu AI build a super-kilobyte scale inference cluster
Since the official launch of "Zhipu Qingyan", it has attracted millions of users every day, facing the problem of text, Large-scale real-time reasoning requirements in multiple scenarios such as pictures and videos. In order to meet the surge in model computing needs, it is necessary to continue to expand the number of computing cards and build a kilo-card level inference cluster to further improve computing resource utilization and inference performance.
Uked’s inference service platform provides ultra-large-scale integrated computing power and supports unified scheduling and management of computing clusters. At present, Ucadex has successfully assisted Zhipu AI in building an inference cluster with a scale of over 1,000 cards. At the same time, with the support of Youked cloud interoperability products, the platform also has powerful "hybrid networking capabilities", allowing large models to achieve integrated training and promotion. Computing resource management based on the full life cycle not only ensures the efficient and stable operation of large models, enabling them to cope with various complex reasoning tasks, but also provides solid technical guarantee for real-time response of cloud services.
Match full-stack computing resources to achieve diversified reasoning scenario coverage
The smart spectrum large model is widely used in intelligent programming, intelligent writing and other fields, providing services for various industries Intelligent upgrades provide strong technical support. Whether processing multi-modal data such as text, images or videos, the smart spectrum large model can demonstrate excellent performance and flexibility.
Uked’s inference service platform matches full-stack computing resources, is compatible with diverse scenarios such as general large models and industry large models, and provides flexible and stable inference services for various models such as text and image generation and code generation. Meet the needs of large-scale real-time reasoning in various computing power scenarios. Among them, "CodeGeeX" is a large-model-based intelligent programming assistant launched by Zhipu AI with the support of Youkede's flexible and flexible computing power deployment solution. It can generate and complete code, automatically add comments, Functions such as code translation and intelligent question and answer help programmers write 20 million lines of code every day, significantly improving work efficiency.
In addition to model inference services on the public cloud, Ucade also supports privatized deployment of large models. Ucade and Zhipu AI are exploring a new way of cooperation based on the "large model all-in-one machine". The jointly launched industry large model solution can better help finance, medical, automobile, manufacturing and other industries quickly implement large model business . At present, Ucade's reasoning service platform has integrated rich industry model resources. These industry models can be customized for different industry needs, providing more accurate and efficient reasoning capabilities.
Significantly reduce inference costs and achieve a balance between cost-effectiveness and service quality
As AIGC technology continues to evolve, its reliance on GPU computing power has become increasingly obvious. While large model companies are pursuing excellent computing performance, they are also paying more and more attention to the utilization efficiency and cost requirements of inference computing power.
Currently, Ukede has introduced advanced GPU resource management and scheduling mechanisms to provide flexible and reliable performance support for large smart spectrum models. Through intelligent allocation and dynamic adjustment of cluster tasks, the load pressure on a single node is effectively reduced, while idleness and excessive consumption of computing resources are avoided. Under this refined resource management method, Ukerde helps significantly improve the computing power utilization of large smart spectrum models, bringing an economical and efficient large model inference experience. Ucade's products are significantly better than similar competitors in terms of inference costs, successfully achieving a balance between cost-effectiveness and service quality.
At the same time, Zhipu AI uses the UPFS parallel file system independently developed by Ukede to optimize model inference performance. UPFS supports IB/RoCE networks, providing access to data in hundreds of microseconds and read and write throughputs of up to hundreds of GB/s, further improving the efficiency of data transmission and communication.
In the future, Ucade will work hand in hand with Zhipu AI to promote the continuous innovation and application of large model technology with a more flexible and reliable intelligent computing base. It is believed that through the close cooperation and unremitting efforts of both parties, large models will take root in various fields and be fully integrated into production and life. More users and more families can enjoy intelligent, efficient and convenient artificial intelligence experiences.
The above is the detailed content of Youked builds a kilo-calorie inference cluster for Zhipu AI to help global users enjoy large-model smart life. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undress AI Tool
Undress images for free

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Clothoff.io
AI clothes remover

Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

But maybe he can’t defeat the old man in the park? The Paris Olympic Games are in full swing, and table tennis has attracted much attention. At the same time, robots have also made new breakthroughs in playing table tennis. Just now, DeepMind proposed the first learning robot agent that can reach the level of human amateur players in competitive table tennis. Paper address: https://arxiv.org/pdf/2408.03906 How good is the DeepMind robot at playing table tennis? Probably on par with human amateur players: both forehand and backhand: the opponent uses a variety of playing styles, and the robot can also withstand: receiving serves with different spins: However, the intensity of the game does not seem to be as intense as the old man in the park. For robots, table tennis

On August 21, the 2024 World Robot Conference was grandly held in Beijing. SenseTime's home robot brand "Yuanluobot SenseRobot" has unveiled its entire family of products, and recently released the Yuanluobot AI chess-playing robot - Chess Professional Edition (hereinafter referred to as "Yuanluobot SenseRobot"), becoming the world's first A chess robot for the home. As the third chess-playing robot product of Yuanluobo, the new Guoxiang robot has undergone a large number of special technical upgrades and innovations in AI and engineering machinery. For the first time, it has realized the ability to pick up three-dimensional chess pieces through mechanical claws on a home robot, and perform human-machine Functions such as chess playing, everyone playing chess, notation review, etc.

The start of school is about to begin, and it’s not just the students who are about to start the new semester who should take care of themselves, but also the large AI models. Some time ago, Reddit was filled with netizens complaining that Claude was getting lazy. "Its level has dropped a lot, it often pauses, and even the output becomes very short. In the first week of release, it could translate a full 4-page document at once, but now it can't even output half a page!" https:// www.reddit.com/r/ClaudeAI/comments/1by8rw8/something_just_feels_wrong_with_claude_in_the/ in a post titled "Totally disappointed with Claude", full of

Deep integration of vision and robot learning. When two robot hands work together smoothly to fold clothes, pour tea, and pack shoes, coupled with the 1X humanoid robot NEO that has been making headlines recently, you may have a feeling: we seem to be entering the age of robots. In fact, these silky movements are the product of advanced robotic technology + exquisite frame design + multi-modal large models. We know that useful robots often require complex and exquisite interactions with the environment, and the environment can be represented as constraints in the spatial and temporal domains. For example, if you want a robot to pour tea, the robot first needs to grasp the handle of the teapot and keep it upright without spilling the tea, then move it smoothly until the mouth of the pot is aligned with the mouth of the cup, and then tilt the teapot at a certain angle. . this

Conference Introduction With the rapid development of science and technology, artificial intelligence has become an important force in promoting social progress. In this era, we are fortunate to witness and participate in the innovation and application of Distributed Artificial Intelligence (DAI). Distributed artificial intelligence is an important branch of the field of artificial intelligence, which has attracted more and more attention in recent years. Agents based on large language models (LLM) have suddenly emerged. By combining the powerful language understanding and generation capabilities of large models, they have shown great potential in natural language interaction, knowledge reasoning, task planning, etc. AIAgent is taking over the big language model and has become a hot topic in the current AI circle. Au

This afternoon, Hongmeng Zhixing officially welcomed new brands and new cars. On August 6, Huawei held the Hongmeng Smart Xingxing S9 and Huawei full-scenario new product launch conference, bringing the panoramic smart flagship sedan Xiangjie S9, the new M7Pro and Huawei novaFlip, MatePad Pro 12.2 inches, the new MatePad Air, Huawei Bisheng With many new all-scenario smart products including the laser printer X1 series, FreeBuds6i, WATCHFIT3 and smart screen S5Pro, from smart travel, smart office to smart wear, Huawei continues to build a full-scenario smart ecosystem to bring consumers a smart experience of the Internet of Everything. Hongmeng Zhixing: In-depth empowerment to promote the upgrading of the smart car industry Huawei joins hands with Chinese automotive industry partners to provide

At this ACL conference, contributors have gained a lot. The six-day ACL2024 is being held in Bangkok, Thailand. ACL is the top international conference in the field of computational linguistics and natural language processing. It is organized by the International Association for Computational Linguistics and is held annually. ACL has always ranked first in academic influence in the field of NLP, and it is also a CCF-A recommended conference. This year's ACL conference is the 62nd and has received more than 400 cutting-edge works in the field of NLP. Yesterday afternoon, the conference announced the best paper and other awards. This time, there are 7 Best Paper Awards (two unpublished), 1 Best Theme Paper Award, and 35 Outstanding Paper Awards. The conference also awarded 3 Resource Paper Awards (ResourceAward) and Social Impact Award (

At the World Robot Conference being held in Beijing, the display of humanoid robots has become the absolute focus of the scene. At the Stardust Intelligent booth, the AI robot assistant S1 performed three major performances of dulcimer, martial arts, and calligraphy in one exhibition area, capable of both literary and martial arts. , attracted a large number of professional audiences and media. The elegant playing on the elastic strings allows the S1 to demonstrate fine operation and absolute control with speed, strength and precision. CCTV News conducted a special report on the imitation learning and intelligent control behind "Calligraphy". Company founder Lai Jie explained that behind the silky movements, the hardware side pursues the best force control and the most human-like body indicators (speed, load) etc.), but on the AI side, the real movement data of people is collected, allowing the robot to become stronger when it encounters a strong situation and learn to evolve quickly. And agile
