Search AI Products and News

With the development of digital art, automated image processing technology has received increasing attention. Recently, a research team from Tsinghua University and Tencent's ARC Lab proposed a novel image sequence coloring model called ColorFlow. This model aims to address the challenge of maintaining consistency in character and object identity while coloring black and white image sequences, fulfilling the practical needs of industries such as comics and animation. ColorFlow is a three-stage diffusion base framework that leverages contextual information by utilizing a reference image pool for black and white.

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.

Recently, researchers from ByteDance Research Institute and Tsinghua University jointly released a new study, pointing out that current AI video generation models, such as OpenAI's Sora, while capable of creating stunning visual effects, have significant flaws in understanding basic physical laws. This study has sparked widespread discussion on the capabilities of AI in simulating reality. The research team tested AI video generation models under three different scenarios: predictions under known patterns and predictions under unknown patterns.

The research team at Tsinghua University recently released a mobile sound source simulation platform named SonicSim, aiming to address the current data scarcity in the field of voice processing for mobile sound source scenarios. This platform is built on the Habitat-sim simulation platform, capable of realistically simulating real-world acoustic environments, providing higher quality data support for the training and evaluation of voice separation and enhancement models. Existing voice separation and enhancement datasets are mostly based on static sound sources, which are inadequate for the needs of mobile sound source scenarios.

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.

In today's rapidly advancing world of artificial intelligence, voice synthesis and transformation technologies are evolving at an extraordinary pace, providing us with incredibly realistic and natural audio experiences. However, these technological advancements also bring potential security risks, particularly as 'voice cloning' technology could be exploited by malicious individuals, threatening personal privacy and social stability. In response to this challenge, the Intelligent System Security Laboratory of Zhejiang University and Tsinghua University have joined forces to launch a revolutionary audio forgery detection framework - SafeEar. This framework not only efficiently detects forged audio but also ensures privacy protection during the detection process.

In the context of rapid advancements in voice synthesis technology, voice forgery has become increasingly severe, posing significant threats to user privacy and social security. Recently, the Intelligent System Security Laboratory at Zhejiang University and Tsinghua University jointly released a novel voice forgery detection framework called 'SafeEar.' This framework aims to efficiently detect forgeries while protecting the privacy of voice content. The concept behind SafeEar is to cleverly design a decoupled model based on neural audio codecs that effectively separates the acoustic features of voice.

In the field of artificial intelligence, an innovative research from China is gaining widespread attention. Scientists from Tsinghua University and the Shanghai Artificial Intelligence Laboratory have proposed a novel framework called 'Diagram of Thought' (DoT), which is expected to revolutionize our understanding of AI thinking patterns. The core idea of the DoT framework is to mimic the human thought process in solving complex problems. Just like how we continually hypothesize, critique, and revise when tackling difficult problems, ultimately arriving at a conclusion, DoT...

The report 'How Innovative is China in the Field of AI?' provides a comprehensive assessment of the development status of China's artificial intelligence sector, offering a comparison with the United States. It highlights that China has become a leader in global AI research publications and is on par with the US in the field of generative AI. China ranks first in the world for the number of papers published in this area and leads in patent numbers over the US. In terms of talent cultivation, Tsinghua University is seen as a breeding ground for AI startups, demonstrating China's leading position in top AI talent. The technical strength of China's large models is rapidly improving, narrowing the performance gap with the US.

The research team at Tsinghua University has successfully developed the 'LongWriter' artificial intelligence system, which can generate coherent texts exceeding 10,000 words, paving new avenues for long-form writing and content creation. By constructing the 'LongWriter-6k' dataset, which includes 6,000 writing samples, the research team has increased the maximum output length of the AI model from approximately 2,000 words to over 10,000 words. This innovative achievement not only enhances the AI's performance in long text generation tasks but even surpasses larger-scale models.

Tsinghua University's Departments of Electronic Engineering and Automation have jointly developed the "Taiji-II" optical training chip, enabling efficient and precise optical computation training for large-scale neural networks. This innovation overcomes the limitations of existing optical computing technology, eliminating the reliance on GPUs for offline modeling and simplifying the training process. The chip demonstrates exceptional performance....

Another significant step has been taken in the field of code generation as the Knowledge Engineering Group (KEG) at Tsinghua University and the Data Mining team have launched their latest innovative achievement: CodeGeeX4-ALL-9B. This model is part of the renowned CodeGeeX series and represents the pinnacle of multilingual code generation, setting new performance and efficiency standards for automated coding. The CodeGeeX4-ALL-9B model is a product of extensive training on the GLM-4-9B framework

In the global field of artificial intelligence research, China is rising at an astonishing speed and quality. According to the latest survey report from the George Washington University data analysis platform ETO, between 2017 and 2022, China published 243,000 AI papers, accounting for 25% of the global total and ranking first in the world. This achievement marks a major breakthrough for China in terms of the quality of AI talent and technological innovation capabilities. The United States follo