News

Huang Renxun revealed the launch time of Blackwell Ultra and Vera Rubin at GTC 2025

Huang Renxun open-sourced a series of new achievements at the GTC 2025 conference, including the Physical AI dataset and the GR00T N1 basic model.

A research team from the University of Cambridge proposed the Celcomen virtual tissue model to unlock the secrets of intracellular and intercellular gene regulation in spatial transcriptomics and single-cell data.

3.10-3.14
Weekly Highlights!

Laoshan Laboratory and others have developed a large-scale intelligent forecasting model for the ocean environment called "Wenhai" through the deep integration of physical oceanography and artificial intelligence, with better performance.

A joint team from Tsinghua University and other universities proposed the CoPRA model, which set a new record in protein-RNA binding affinity prediction capabilities.

Contains OpenManus + QwQ-32B one-click deployment tutorial.

David Baker's team combined fine-tuned RFdiffusion with yeast display screening to demonstrate the feasibility of de novo antibody domain design

Scientists express concern about Musk's reliance on AI tools to determine the scale of layoffs and budget cuts in government agencies

3.3-3.7
Weekly Highlights!

Contains QwQ-32B one-click deployment tutorial.

Tsinghua University and the Chinese Academy of Sciences team proposed the UniGEM model, which for the first time achieved collaborative enhancement of two tasks based on a diffusion model.

The third session of the 14th National People's Congress was held to summarize the achievements of scientific and technological development in 2024 and provide prospects for future work.

NVIDIA has collaborated with teams from multiple research institutes and universities to develop Proteina, a large-scale flow protein backbone generator that can generate diverse and designable proteins.

Professor Shen Chunhua's team from Zhejiang University, in collaboration with teams from several foreign universities, proposed the Boltzmann alignment technique, which transfers knowledge from the pre-trained inverse folding model to the prediction of ∆∆G.

2.24-2.28
Weekly Highlights

Hyper AI has prepared 7 high-quality reasoning data sets for everyone, covering mathematics/code/science/puzzles and other aspects.

A research team from Zhejiang Provincial GIS Key Laboratory proposed a deep learning model based on attention mechanism to more accurately estimate spatial non-stationarity.

The Australian team proposed the use of an automated framework based on deep learning that can combine gender and skull features for judgment, greatly improving the accuracy of gender judgment.

Contains the Step-Audio-TTS one-click deployment tutorial.

The Korean team introduced the Retrieval-Retro inorganic retrosynthesis planning method, which greatly improved the efficiency and accuracy of inorganic material synthesis.

The 7th episode of the Meet AI4S live series has invited three lecturers in the field of artificial intelligence, including Professor Huang Hong from Huazhong University of Science and Technology. Come and make an appointment for the live broadcast!

Oxford/Amazon/Westlake University/Tencent and others proposed the multimodal, multi-domain, and multi-language medical model M³FM, which supports disease reporting and disease classification, and performs better than previous methods.

The Duke University research team constructed the PepPrCLIP pipeline, which can design short proteins (peptides) that can bind to and destroy previously undruggable disease-causing proteins.

The "Tutorial" section of SuperNeural's official website has launched 3 sound cloning models, which can be deployed and experienced with one click.

A team from the Hong Kong Polytechnic University published the RingFormer framework, which greatly shortened the research and development cycle of new materials and pushed organic solar cell research into a new era.

2.5-2.14
Weekly Highlights

Owkin has built the world's largest cancer spatial omics dataset, solving the patient data privacy issue that the public is most concerned about, and facilitating cancer diagnosis and drug development.

The China University of Geosciences team and Baidu jointly proposed the image segmentation model ConDSeg, which solved the two major problems of "soft boundaries" and co-occurrence in the field of medical image segmentation.

Contains one-click deployment DeepSeek-R1-70B tutorial