Privacy Filter is an open-source bidirectional labeling and classification model developed by OpenAI for cleaning high-throughput data. It is used to efficiently detect and mask personally identifiable information (PII) in text. It is based on a small pre-trained architecture similar to gpt-oss and abandons the traditional word-by-word generation method. Instead, it directly decodes coherent segments of the input sequence through a single forward propagation combined with the constrained Viterbi algorithm.

Currently, the HyperAI website has launched [the relevant section/feature].Privacy Filter ModelCome and try it!

Online use:https://go.hyper.ai/Py1l3

A quick overview of hyper.ai's official website updates from April 25th to April 30th:

* High-quality public datasets: 5

* A selection of high-quality tutorials: 5

* Community article analysis: 1 article

* Popular encyclopedia entries: 5

Visit the official website:hyper.ai

Selected public datasets

1. Transfermarkt Football dataset

Transfermarkt Football is a structured football transfer market dataset built on the Transfermark website, designed for sports analytics and data modeling. The dataset contains over 80,000 football matches, 400 clubs, and more than 37,000 players, recording player market value changes, appearances, and transfer activity.

Online use:https://go.hyper.ai/lF661

2. Yoga Training: Yoga posture classification and training dataset

Yoga Training is a dataset for yoga posture classification, primarily used for image classification, posture recognition, lightweight deep learning training, and transfer learning experiments. This dataset contains 1,771 sample images of yoga postures, covering a wide range of difficulty levels and posture categories.

Online use:https://go.hyper.ai/hVdM8

3. Corn Leaf Diseases Dataset

Corn Leaf Diseases is a dataset of corn leaf images specifically designed for target detection tasks in precision agriculture. The dataset contains 4,027 corn leaf images, covering four categories, including healthy corn leaves and three common diseases: rust, gray spot, and wilt.

Online use:https://go.hyper.ai/UbRRp

4. Apple Leaf Diseases Dataset

Apple Leaf Diseases is a high-quality apple leaf image dataset specifically designed for target detection tasks in precision agriculture. The dataset contains 3,444 apple leaf images, covering four categories: healthy apple leaves and three common diseases: black rot, cedar rust, and scab.

Online use:https://go.hyper.ai/LDafw

5. Drug Adverse Event Detection Dataset

Drug Adverse Event Detection is a text dataset that simulates real-world scenarios involving multiple drug prescriptions for patients. It aims to study the risk of adverse drug reactions caused by the combined use of multiple drugs and has wide applications in adverse drug reaction detection, medical information extraction, clinical text analysis, and medical AI model training.

Online use:https://go.hyper.ai/AlL32

Selected Public Tutorials

1. Privacy Filter Model

OpenAI Privacy Filter is a bidirectional token classification model released by OpenAI in April 2026, used to detect and mask personally identifiable information (PII) in text. The model adopts an architecture similar to gpt-oss but smaller in scale. The official model card description states that it has approximately 1.5B total parameters, approximately 50M active parameters, supports a maximum of 128K token context, and outputs privacy fragment boundaries through 33 BIOES token-level labels.

Run online:https://go.hyper.ai/Py1l3

2. Hermes Operating Tutorial

Hermes Agent is an open-source, self-evolving AI agent developed by the Nous Research team in 2026. A core feature of this project is its built-in learning loop—it automatically creates skills from task experience, continuously improves during use, proactively persists knowledge to its memory system, and can search historical conversations to gradually build a deep understanding of the user across conversations. This website provides tutorials for running Hermes on both GPU and CPU.

Running GPU version online:https://go.hyper.ai/nnyFT

Online CPU version:https://go.hyper.ai/kdo9i

3. One-click deployment of DeepSeek-V4-Flash

DeepSeek V4 is the latest generation of large language models released by the DeepSeek team, including two versions: DeepSeek-V4-Pro (1.6T parameters) and DeepSeek-V4-Flash (285B parameters). DeepSeek V4 adopts a brand-new, highly efficient long context attention mechanism, natively supporting context lengths of up to 1 million tokens, and is designed specifically for handling ultra-long text tasks.

Run online:https://go.hyper.ai/sFyxU

4. Deploy MOSS-TTS-Nano using Free-CPU

MOSS-TTS-Nano is a 0.1B parameter multilingual text-to-speech model released by the OpenMOSS team in April 2026. It supports speech generation and cloning in a CPU environment. The model is designed to balance the naturalness of text-to-speech generation, cross-language usability, and reference audio-driven timbre transfer capabilities, enabling it to cover a variety of common tasks from basic reading aloud to speech cloning.

Run online:https://go.hyper.ai/CwMEH

Community article interpretation

1. Using stacked ensemble learning, a UK research team has achieved high-precision prediction of the seismic index of 251 Delta Scuti stars.

A research team at the University of Warwick in the UK has developed a stacked ensemble learning framework to directly predict key asteroseismic parameters of Delta Scuti stars from TESS light curves. This method achieved remarkable results on a sample of 643 stars: the coefficient of determination (R²) for all target parameters was higher than 0.77, and it demonstrated good generalization ability on 60 stars not used in the training. The prediction results were highly consistent with traditional asteroseismic analysis.

View the full report:https://go.hyper.ai/mNGlM

Popular Encyclopedia Articles

1. Skills

2. HyperNetworks

3. Sigmoid Function

4. Reciprocal Rank Fusion

5. Multi-Agent Architecture

Here are hundreds of AI-related terms compiled to help you understand "artificial intelligence" here:

https://go.hyper.ai/wiki

The above is all the content of this week’s editor’s selection. If you have resources that you want to include on the hyper.ai official website, you are also welcome to leave a message or submit an article to tell us!

See you next week!

About HyperAI

HyperAI (hyper.ai) is the leading artificial intelligence and high-performance computing community in China.We are committed to becoming the infrastructure in the field of data science in China and providing rich and high-quality public resources for domestic developers. So far, we have:

* Provides domestic accelerated download nodes for 2100+ public datasets

* Includes 700+ classic and popular online tutorials

* Analyzing 300+ AI4Science Paper Cases

* Supports searching for 700+ related terms

* Hosting the first complete Apache TVM Chinese documentation in China

Visit the official website to start your learning journey:

https://hyper.ai