Google launched its smartest model yet, the Gemini 2.5 Pro, in March

A team from MIT and Harvard University proposed a framework called PUPS that combines protein sequences and cell images to predict the subcellular localization of unknown proteins.

Qi Jin, a full-time researcher at the School of Earth Sciences of Zhejiang University, gave a special presentation on the "Deep Time Earth Crowd Intelligence Collaborative Innovation Platform".

"ACE-Step: Basic Model for Music Generation" can synthesize up to 4 minutes of music in just 20 seconds, which is 15 times faster than the baseline method based on LLM

Huang Renxun shared several updates about Nvidia in the fields of data center, enterprise AI and robotics at Computex 2025.

Cornell University and Regeneron Pharmaceuticals in the United States proposed a graph-encoded mixed survival model (GEMS) to identify sub-phenotypes with consistent characteristics and survival outcomes.

HyperAI has compiled some of the most popular medical data sets for you, covering medical question-and-answer, medical reasoning, medical imaging and other data.

Google DeepMind announced the programming AI Agent AlphaEvolve, which can be used for general algorithm discovery and optimization.

Researchers from Columbia University and Stanford University proposed a generative artificial intelligence structure analysis method PXRDnet based on a diffusion model.

"In-Context Edit: Command-driven Image Generation and Editing" has been launched in the "Tutorial" section of HyperAI's official website. Only very few text commands are needed to achieve accurate image modification. Come and experience it!

David Baker's team at the University of Washington recently used advanced generative models to conduct synthetic OLG design research and verify its feasibility from an engineering perspective.

Researchers from the Russian Academy of Sciences have developed a machine learning-based search engine, MEDUSA Search, that can analyze terabyte-scale high-resolution mass spectrometry data to help discover unknown chemical reactions!

10 practical tutorials and model examples to quickly master vLLM

Achieve segmentation and generalization capabilities that are superior to existing advanced models in zero-shot, one-shot, and few-shot scenarios.

The team proposed an innovative framework to reveal the unique "two-step" ion migration mechanism in hydride SSEs

The HIT team proposed a hierarchical distillation multi-instance learning framework HDMIL, which can significantly reduce the inference time

Contains Qwen3 one-click deployment tutorial

The team led by Zhou Hao from Tsinghua University AIR proposed ProfileBFN (Profile Bayesian Flow Network), which achieves efficient protein family design

IBM Research and others jointly launch EarthDial to provide strong support for earth observation data

A team from Waseda University in Japan used machine learning technology to perform molecular design and experimental optimization of light-driven crystals to successfully maximize the blocking force

4.21-4.25
Weekly Highlights!

Contains 12 HPC tutorials summary

MindGlide is expected to further improve the ability of medical staff to interpret and evaluate the effectiveness of treatment for MS patients

This method uses a one-class support vector machine (one-class SVM) to analyze the characteristic differences between niacinamide and niacin in UV spectra.

UNO is based on the FLUX model and can handle different input conditions in image generation tasks.

The DRAKES algorithm implements for the first time differentiable reward backpropagation for fully generated trajectories in a discrete diffusion model.

M2OST is a many-to-one regression Transformer model designed to jointly predict gene expression using different levels of pathological images.

HyperAI has compiled multiple material science data sets and one-click deployment tutorials, covering key areas such as quantum materials, inorganic materials, and crystal structures.

The protein engineering design platform VenusFactory achieves one-stop optimization of data retrieval/model training/benchmark evaluation, lowering the threshold for AI4S applications.

Contains DeepCoder-14B-Preview one-click deployment tutorial