Currently, more and more enterprises are applying AI to their business, but the process is also plagued by issues such as lack of professional skills, insufficient data, and long development cycles. To address this issue, NVIDIA has launched a new TAO "Dao" framework to simplify AI workflows, achieving the effect of shortening time and reducing costs. Over the past five years, NVIDIA has invested hundreds of millions of GPU computing hours to improve and debug thousands of pre trained models (PTM, including computer vision, conversational AI, and even natural language understanding). These high-precision models are detailed on NGC (NVIDIA GPU Cloud) and are available for free download and use. TAO provides the transfer learning toolkit (TLT), which enables developers to use existing data sets locally, simplify data preparation process, speed up model training, and get their own models more quickly. TAO also supports federated learning to ensure secure collaboration among developers from different locations. TAO provides the TensorRT module, which adjusts the parameters of the model based on the target system it is running on to achieve the optimal balance of minimum scale and maximum accuracy. At the same time, it uses the Triton Inference Server, allowing users to choose the optimal configuration to deploy any architecture and framework model to the target CPU and GPU. The Avenue to Simplify, NVIDIA TAO has simplified the AI workflow and built a fast lane towards corporate AI.

Live highlights:

-NVIDIA TAO simplifies AI workflow and builds a fast lane towards enterprise AI

-NVIDIA Pre trained Models, Transfer Learning Toolkit 3.0, Deepstream 6.0, TensorRT 8.0, etc

Live dry goods:

This Webnair will introduce the modules and related cases of NVIDIA TAO in detail, including the pre trained model on NVIDIA GPU Cloud (NGC), the transfer learning toolkit, the new reasoning optimization and deployment tools TensorRT 8.0 and Triton.

Suitable audience:

Developers and managers who focus on the rapid implementation of artificial intelligence algorithms and applications.