Hello,
We have a Jetson Nano 4GB, and would like to run code that involves both Tensorflow and PyTorch on it.
After installing all cuda drivers, we don’t have enough space to install both tensorflow and Pytorch. Is there a solution to this problem?
In the event that we have to rewrite our code and model, which one would be preferred - Tensorflow or PyTorch?
Do you recommend working on those Jetsons with NVIDIA Docker or directly on the system?
Hi @maayan2, you might be interested in this guide from @cyato about Minimizing Disk Usage on Jetson: https://nvidia-ai-iot.github.io/jetson-min-disk/
If possible, you may want to attach external storage for additional disk space. I personally use Docker for ML frameworks like PyTorch and TensorFlow - the l4t-ml container comes with both PyTorch and TensorFlow pre-installed (in addition to other ML-related packages, so it is a larger container and you may want to customize it’s Dockerfile)
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.