Skip to main content

ReleaseNote v1.0

1. Embodied Intelligence Zone​

  • Introduced the "Well-known Enterprises" module. Users can now explore leading companies in the field of embodied intelligence and stay updated on industry trends and technological advancements.

2. DataFlow Platform Launched​

  • The DataFlow platform, built on Agentic architecture, offers dynamic and highly customizable solutions for complex data processing tasks. It is designed to provide a seamless, all-in-one data processing experience for users. Feel free to contact us for more details and personalized consultation.

3. Support for NVIDIA Inference Engine NIM​

  • NVIDIA’s NIM microservices include pre-built containers for Triton Inference Server™ and TensorRT™-LLM, offering powerful inference acceleration. With support for high-performance GPUs like A100 and H800, NIM reduces deployment time from weeks to minutes, enabling users to optimize and accelerate model inference with ease.

4. Support for Ascend Inference Engine MINDIE​

  • The Ascend Inference Engine MindIE is an inference acceleration suite designed for AI across various scenarios. By providing layered AI capabilities, it meets the diverse needs of users while maximizing the computational power of Ascend hardware. MindIE supports various types of Ascend AI processors and offers multi-level programming interfaces, enabling users to quickly develop AI applications on the Ascend platform.

5. Prompt Engineering Zone​

  • We’ve launched a new Prompt Engineering Zone for the community, offering tools and resources related to prompt engineering.
  • In the “Prompt Library” module, users can browse and share high-quality prompts.
  • The “Optimization Optimizer" module provides a model service specifically designed to optimize prompts based on the user’s needs and scenarios.

6. Dataset Preview Feature Enhancement​

  • Added support for recognizing and switching between subsets and usage-specific datasets for Parquet format datasets.

7. Backend Storage System Upgrade​

  • Our backend storage system has undergone a major upgrade with Gitaly-based model synchronization. This enhancement significantly improves system availability, response speed, and stability.