Using TLT to fine-tune models on Jetson devices

Hello community!

I am relatively new to Jetson development processes and I would like to fine-tune a model from the ones available on TLT. However, there is a super simple question that I couldn’t find a concrete answer for on my searches.

Using TLT, can i actually perform the training (fine-tuning) process on a Jetson Nano or Jetson Xavier NX?

I remember seeing somewhere when I started using the Jetson that deep learning model training could not be performed on Jetson devices, and had to be performed on a different platform, such as a “normal” x86/64 PC

The training/pruning/retraining are designed to run on x86 systems with an NVIDIA GPU (e.g., GPU-powered workstation, DGX system) or can be run in any cloud with an NVIDIA GPU.

https://docs.nvidia.com/metropolis/TLT/tlt-user-guide/text/requirements_and_installation.html#hardware-requirements

The following system configuration is recommended to achieve reasonable training performance with the TLT and supported models provided:

  • 32 GB system RAM
  • 32 GB of GPU RAM
  • 8 core CPU
  • 1 NVIDIA GPU
  • 100 GB of SSD space

Thank you for your feedback!