A research team from the University of Pennsylvania in the United States integrated four major venom databases to build a global venom database and applied a sequence-to-function deep learning model called APEX.

"OCRFlux-3B: Intelligent Text Recognition Toolkit" has been launched on the tutorial section of HyperAI's official website. You can experience it online with one-click deployment~

Universiti Putra Malaysia and the University of New South Wales Sydney jointly proposed an intelligent framework for the automatic classification of ceramic artifacts and the estimation of their market value.

Peking University and OpenBayes jointly launched the text-based panoramic video generation framework PanoWan.

HyperAI has compiled a series of valuable and widely used tutorials and datasets for everyone from July 21 to July 25, covering multiple fields such as audio analysis and large models~

The 7th 2025 Meet AI Compiler Technology Salon concluded successfully on July 5 in Zhongguancun, Beijing.

Aeneas breaks through the limitations of repair and creates the ability of "repair of arbitrary length" for the first time.

July 14-July 18 AI Paper Selection

Hunan University, in collaboration with the University of Chinese Academy of Sciences and other teams, proposed APM, a model that surpasses existing SOTA performance in downstream tasks such as antibody design and binding peptide design.

The Apache Software Foundation (ASF)'s official global series of conferences, Community Over Code Asia 2025, will be held in Beijing from July 25 to 27.

Dr. Wang Lei, the founder of the TileAI community, gave a thorough and easy-to-understand introduction to the innovative operator programming language TileLang. This article is a transcript of his in-depth sharing.

On July 25, Apache will hold the Community Over Code Asia 2025 AI Forum in Zhongguancun, Beijing, bringing the latest information on Apache community building and development.

"Voxtral-Mini-3B-2507 speech understanding model demo" is now available on the "Tutorial" section of HyperAI's official website (hyper.ai). Come and try it out~

The group of Professor Zheng Yinqiang from the University of Tokyo and the group of Professor Ding Jun from McGill University jointly proposed a method SUICA for modeling spatial transcriptome data.

HyperAI has compiled a series of valuable and widely used tutorials and datasets for everyone from July 14 to July 18, covering large language models, agriculture and other fields~

Zheng Size, a Seed Research Scientist from ByteDance, analyzed in detail the communication efficiency breakthrough and cross-platform adaptability of Triton-distributed in large-model training.

The NVIDIA research team and Mila proposed La-Proteina, an atomic-level protein design method based on partial latent flow matching.

July 7-July 11 AI Paper Selection

At the 7th Meet AI Compiler Technology Salon hosted by HyperAI, Dong Zhaohua, senior director of Muxi Integrated Circuit, gave an in-depth sharing.

"Biomni: The First Universal Biomedical Intelligent Agent" is now available on the "Tutorial" section of HyperAI's official website. Come and experience it!

The new protein sequence design method FAMPNN can process protein main chain and side chain information simultaneously, significantly improving the quality of protein sequence design and the accuracy of experimental prediction

7.7-7.11
Weekly Highlights

Atomic Diffusion Transformer ADiT, the first single model to generate molecules and crystals

FLUX.1 Kontext is now open source, including a one-click deployment tutorial

A selection of AI papers from June 30th to July 4th, including download links for high-quality datasets~

The 7th 2025 Meet AI Compiler Technology Salon concluded successfully on July 5 in Zhongguancun, Beijing.

The National University of Singapore and Zhejiang University jointly proposed the innovative NeuralCohort method, opening up a new path for EHR representation learning.

HyperAI has compiled a series of extremely valuable and widely used datasets and tutorials for everyone from June 30 to July 4, covering multiple fields such as biological vision, medical analysis, and agriculture~

Specializing in AI review? Papers contain hidden positive feedback, Xie Saining calls for attention to the evolution of scientific research ethics in the AI era

REVERIE is the world's first virtual reality intelligent exercise intervention system for weight control in overweight or obese adolescents.

A research team from the University of Pennsylvania in the United States integrated four major venom databases to build a global venom database and applied a sequence-to-function deep learning model called APEX.

"OCRFlux-3B: Intelligent Text Recognition Toolkit" has been launched on the tutorial section of HyperAI's official website. You can experience it online with one-click deployment~

Universiti Putra Malaysia and the University of New South Wales Sydney jointly proposed an intelligent framework for the automatic classification of ceramic artifacts and the estimation of their market value.

Peking University and OpenBayes jointly launched the text-based panoramic video generation framework PanoWan.

HyperAI has compiled a series of valuable and widely used tutorials and datasets for everyone from July 21 to July 25, covering multiple fields such as audio analysis and large models~

The 7th 2025 Meet AI Compiler Technology Salon concluded successfully on July 5 in Zhongguancun, Beijing.

Aeneas breaks through the limitations of repair and creates the ability of "repair of arbitrary length" for the first time.

July 14-July 18 AI Paper Selection

Hunan University, in collaboration with the University of Chinese Academy of Sciences and other teams, proposed APM, a model that surpasses existing SOTA performance in downstream tasks such as antibody design and binding peptide design.

The Apache Software Foundation (ASF)'s official global series of conferences, Community Over Code Asia 2025, will be held in Beijing from July 25 to 27.

Dr. Wang Lei, the founder of the TileAI community, gave a thorough and easy-to-understand introduction to the innovative operator programming language TileLang. This article is a transcript of his in-depth sharing.

On July 25, Apache will hold the Community Over Code Asia 2025 AI Forum in Zhongguancun, Beijing, bringing the latest information on Apache community building and development.

"Voxtral-Mini-3B-2507 speech understanding model demo" is now available on the "Tutorial" section of HyperAI's official website (hyper.ai). Come and try it out~

The group of Professor Zheng Yinqiang from the University of Tokyo and the group of Professor Ding Jun from McGill University jointly proposed a method SUICA for modeling spatial transcriptome data.

HyperAI has compiled a series of valuable and widely used tutorials and datasets for everyone from July 14 to July 18, covering large language models, agriculture and other fields~

Zheng Size, a Seed Research Scientist from ByteDance, analyzed in detail the communication efficiency breakthrough and cross-platform adaptability of Triton-distributed in large-model training.

The NVIDIA research team and Mila proposed La-Proteina, an atomic-level protein design method based on partial latent flow matching.

July 7-July 11 AI Paper Selection

At the 7th Meet AI Compiler Technology Salon hosted by HyperAI, Dong Zhaohua, senior director of Muxi Integrated Circuit, gave an in-depth sharing.

"Biomni: The First Universal Biomedical Intelligent Agent" is now available on the "Tutorial" section of HyperAI's official website. Come and experience it!

The new protein sequence design method FAMPNN can process protein main chain and side chain information simultaneously, significantly improving the quality of protein sequence design and the accuracy of experimental prediction

7.7-7.11
Weekly Highlights

Atomic Diffusion Transformer ADiT, the first single model to generate molecules and crystals

FLUX.1 Kontext is now open source, including a one-click deployment tutorial

A selection of AI papers from June 30th to July 4th, including download links for high-quality datasets~

The 7th 2025 Meet AI Compiler Technology Salon concluded successfully on July 5 in Zhongguancun, Beijing.

The National University of Singapore and Zhejiang University jointly proposed the innovative NeuralCohort method, opening up a new path for EHR representation learning.

HyperAI has compiled a series of extremely valuable and widely used datasets and tutorials for everyone from June 30 to July 4, covering multiple fields such as biological vision, medical analysis, and agriculture~

Specializing in AI review? Papers contain hidden positive feedback, Xie Saining calls for attention to the evolution of scientific research ethics in the AI era

REVERIE is the world's first virtual reality intelligent exercise intervention system for weight control in overweight or obese adolescents.