News

Speed Equals Intelligence! TileRT Team Member Ma Lingxiao Explains: Computational Exploration and Collaborative Design for ultra-low Latency Large Model inference.

On August 1st, the 9th Meet AI Compiler Technical Salon, hosted by HyperAI, successfully concluded in Zhongguancun, Beijing. This salon was as popular as ever, with industry experts providing insightful presentations, bringing together representatives from the Beijing Academy of Artificial Intelligence, the TileRT team, Tencent, Huawei Ascend, [...]

12 hours ago

Chen Tianqi: In the Agent Era, How Should We Rethink AI Compilers?

Chen Tianqi shared the current development direction of the TVM open source community and had in-depth exchanges with the audience.

5 days ago

Online Tutorial | ProteinGym #1! VenusREM Uses "search Enhancement" to Predict the Impact of Protein Mutations and Accelerate Protein design.

VenusREM is now available on the HyperAI tutorial section. Follow the tutorials to complete the entire process from environment configuration and model deployment to mutation effect prediction and result visualization.

5 days ago

Biopharmaceuticals

The Beijing Academy of Artificial Intelligence (BAAI) Proposed the Triton-TLE Layered Language Extension and Achieved a Hundredfold Automatic Tuning Speedup Through FlagTree Compilation optimization.

Guo Hui and Xiao Hang, researchers at BAAI AI Compiler, introduced a series of explorations conducted by the FlagTree team around Triton.

6 days ago

The First Standardized Benchmark for Protein Mutation Ranking in LLM! A Harvard University Team Proposes PG-LLM, Which Evaluates 13 Mainstream Models and 95 Specialized Models in One go.

A research team from Harvard University and Capable proposed a PG-LLM benchmark built on the ProteinGym dataset.

6 days ago

Online Tutorial | Achieving Raw Image Editing With 4B Parameters: Microsoft Open-Sources Mage-Flow, Enabling Second-Level Image Inference and Supporting Up to 2048 Resolution

Mage-Flow is now available on HyperAI (hyper.ai), helping developers experience high-quality image generation and editing at a lower cost. Interested users can explore it with a single click!

7 days ago

The U.S. Geological Survey Has Proposed the DAS Earthquake Early Warning Model, Which Uses 4-second P-wave Strain Data to Rapidly Classify Moderate to Strong earthquakes.

A research team from the U.S. Geological Survey has bypassed the traditional approach of "locating the earthquake first and then estimating its magnitude" and proposed a binary magnitude classification method based on machine learning.

8 days ago

One Day to Go! Zhiyuan, TileRT, Tencent, Huawei, and Zhiyuan Innovation Gather to Explore multi-level Collaborative Optimization of AI compilation.

The 9th Meet AI Compiler event will officially begin tomorrow at 13:30. See you there!

11 days ago

Technology Salon

Activity

MiniCPM-RobotManip, With Its 1.5 Billion Parameters, Is Officially Open-sourced; NVIDIA Releases the full-modal Model Cosmos3-Edge, Covering Text, Images, Videos, and Motion sequences.

7.25-7.30
Weekly Highlights

12 days ago

Dataset Compilation | From Competition Math to Tool Applications: 9 Open-Source Math Datasets From MIT, NVIDIA, Huazhong University of Science and Technology, etc., Covering CoT, Multimodal Reasoning, and Long-Chain Thinking Training

The dataset is now available on HyperAI, helping researchers and developers quickly explore training, evaluation, and applications.

13 days ago

Tool Call

Reinforcement Learning

Online Tutorial | Run Agents Without Billions of Parameters! Boss Zhipin's Nanbeige Lab Open Sources Nanbeige 4.2-3B, Giving Small Models a "Brain"

Nanbeige 4.2-3B is now available on HyperAI (hyper.ai), helping developers explore native intelligent agent applications at a lower cost. Interested users can explore it with one click!

14 days ago

Intelligent Agent

Long Document Processing

Based on Large Model Inference and the Use of the MCP Tool, AI X-ray Scientists at Stanford University Autonomously Completed single-crystal Diffraction Alignment at a Synchrotron Radiation source.

A team from Stanford University and others developed an "AI X-ray scientist" system and deployed it on a real beamline at the Stanford synchrotron radiation source.

15 days ago

Materials Science

Scientific Research

Text/LaTeX/HTML Tables All in One Step! OvisOCR2 Achieves end-to-end Intelligent Document Parsing; 14,000+ Element Annotations, Tens of Thousands of Language Commands! Voxel51 Releases the SceneFun3D Indoor Scene micro-interaction dataset.

7.18-7.24
Weekly Highlights

16 days ago

Online Tutorial | Compress a 27B Large Model to 7.2GB! Ternary-Bonsai Uses "ternary Magic" to Make Large Models Run on Personal computers.

Ternary-Bonsai-27B is now available on HyperAI (hyper.ai), supporting one-click deployment and quick access. Give it a try!

19 days ago

Argonne National Laboratory in the United States Proposed ChemGraph, Which Uses 13 Benchmark Tests to Evaluate the Value of Agents in the Field of Computational chemistry.

A research team at Argonne National Laboratory in the United States has proposed a smart agent system called ChemGraph, driven by a large language model.

19 days ago

LLM

Online Tutorial | Load ComfyUI Workflow With One Click: Master AI Drawing Without Writing a Single Line of Code

The HyperAI (hyper.ai) tutorial section now includes ComfyUI workflow and ComfyUI × Wan 2.2 model resources. Deploy your model with a single click!

21 days ago

Based on Training With Over 10,000 Tumor Samples, Harvard Medical School and Others Proposed COMPASS, a pan-cancer Basic Model, Which Outperforms 22 Existing Methods on average.

A research team from Harvard Medical School, Roche Pharmaceuticals, and Zhejiang University has proposed a pan-cancer basic model called COMPASS.

22 days ago

Google Releases TabFM-1.0.0-PyTorch: a zero-shot Prediction Model Designed for Mixed Tabular Data; NVIDIA open-sources Multinational Synthetic Character Dataset, With Tens of Millions of Characters available.

7.11-7.17
Weekly Highlights

25 days ago

Dataset Summary | NVIDIA Open Sources Nemotron Datasets: Over 10TB of Tokens + 40M Training Samples, Covering Mathematical Reasoning, Code Generation, and Multilingual dialogue.

Training data is becoming a key variable in the competition for large models. When the number of parameters is no longer the sole barrier, the quality, structure, and task suitability of the data begin to determine the model's true performance in inference, coding, and interaction. NVIDIA's Nemotron series of datasets are built precisely to meet this trend [...]

a month ago

Code Generation

Dataset

Virtual Character Portrait

Training Data

RNA Structure Prediction Rivals AlphaFold 3! A Virginia Tech Team Proposes RNAbpFlow, Which Is Completely Independent of Evolutionary information.

A team from Virginia Tech has developed an SE(3) isovariant flow matching model, RNAbpFlow, based on sequence and base pair conditions.

a month ago

Transformer

Early Bird Tickets Countdown! The AI Compiler Technology Salon in Beijing Is Coming!

The 9th Meet AI Compiler technical salon will be held on August 1st in Zhongguancun Science City, Haidian District, Beijing.

a month ago

AI Compilation

Technology Salon

Online Tutorial | Even a Small 9B Model Can Achieve Complex Reasoning: Based on Qwen 3.5-9B, Qwythos Integrates Claude's Reasoning Experience to Achieve a Leap in Capabilities

HyperAI (hyper.ai) has compiled relevant model resources to help developers deploy models. Interested users can try it out with a single click!

a month ago

NLP

Multimodality

Visual Understanding

OCR Tutorial Compilation | Covering Long Documents, End-to-end, and multi-language Scenarios, Featuring high-performance open-source Models From Baidu, Xiaohongshu, Huazhong University of Science and Technology, and Others for Different Scenarios, Enabling Accurate Multimodal Document parsing.

The tutorial section of HyperAI's official website (hyper.ai) has launched several high-quality open-source OCR models.

a month ago

OCR

Multimodality

Multilingual Processing

Document Parsing

Long Document Processing

ICML 26 Outstanding Papers: Tsinghua JustGRPO Overcomes the dLLM Inference Bottleneck; Say Goodbye to Simple Instruction Tests: Agents Last Exam Comprehensively Evaluates the long-range Professional Capabilities of Intelligent agents.

7.04-7.10
Weekly Highlights

a month ago

Multimodal Large Model

Reinforcement Learning

Speech Synthesis

Technical Salon | Zhiyuan, TileRT, Tencent, Huawei, and Zhiyuan Innovation Gather in Beijing to Focus on AI Compilation multi-level Optimization practices.

The 9th Meet AI Compiler technical salon will be held on August 1st in Zhongguancun Science City, Haidian District, Beijing.

a month ago

AI Compilation

HyperAI New Feature Launched | Configuring Environment Variables and Secret Variables for Computing Containers

HyperAI (hyper.ai) has now officially launched the feature to configure "environment variables and secret variables" for computing containers.

a month ago

OpenAI Releases GeneBench-Pro, Which Assesses AI Research Capabilities Across 129 Questions and 10 domains.

The OpenAI research team recently proposed an updated version of GeneBench, GeneBench-Pro, which covers a wider range of industry and academic fields.

a month ago

Dataset

Life Sciences

Dataset Compilation | AI Agent Evaluation Datasets: 10 Datasets Released by Microsoft, Peking University, HKU, Shanghai Jiao Tong University, etc., Covering Everything From long-range Memory to real-world Task execution.

This article compiles 10 datasets related to AI Agent capability assessment, which can be used online and cover different capability areas such as long-range memory, multi-step reasoning, and tool invocation.

a month ago

Can Emojis Control Speech Generation? Irodori-TTS Is a Japanese TTS Based on the RF-DiT Architecture; Eczema and Tinea Skin Disease Datasets: Supporting Medical Image Classification and Transfer learning.

6.27-7.03
Weekly Highlights

a month ago

Machine Vision