Announcing the new version of the NVIDIA TAO Toolkit v3.22.05

Today NVIDIA announced the release of the latest version of TAO Toolkit (3.22.05) . The latest edition packs a lots of amazing features for our developers as wells as opens up new frontier for AI creation and development. Highlight of the new features include :

  • “Bring your own” ONNX models weights into TAO for fine-tuning and optimizing.
  • Deploy TAO Toolkit as-a-Service in a modern, cloud-native infrastructure on Kubernetes
  • Integrate TAO into your apps using REST APIs.
  • Visualize the model training progress such as training loss, validation loss, and histogram of model weights in TensorBoard.
  • Access new vision pretrained models, including Point Cloud, Pose Action Classification, and 3D Pose Estimation.
  • New speech pretrained models such as Conformer for ASR and HiFiGan and FastPitch for custom text-to-speech applications.

Read the dev news

New developer resources

  • Watch demo videos of new features including Rest APIs, TensorBoard, and pretrained models.
  • Dev blog: Develop and deploy AI-powered Robots with NVIDIA Isaac Sim and NVIDIA TAO.
  • Partner blog: Accelerating AI Model Development with Lexset Synthetic Data and NVIDIA TAO.
  • Success story:
  • TAO Toolkit whitepaper : Transform development efforts in a retail action recognition use case