Compatibility with Jetson Nano 4GB

Trying to run an experiment on Jetson Nano (4GB version) with all the requirements installed, but I keep receiving the error ''MemoryError'' on terminal. Is that a hardware problem of the Jetson Nano (inefficient memory)?
Note: I also increased the size of the swap memory (I use a 32GB SD card) , but it didn't solve the problem.

[gante]

Hey @konarva 👋 by "running the experiment", I'm assuming you mean doing inference using a model trained somewhere else :)

I ran the inference experiment long ago, so my ability to help will be limited. The model itself should be much smaller than 4GB, so I don't think it is the issue. The python environment, however, might have changed significantly 🤔 I'd suggest trying to ensure you use the same packages as in requirements.txt from this specific commit, which corresponds to the codebase at the time of the experiment :)

Hello,
I already have installed the requirements, and I'm trying to run the following command lines:

python3 bin/preprocess_dataset.py /path/to/config.yaml
python3 bin/train_model.py /path/to/config.yaml
python3 bin/test_model.py /path/to/config.yaml
python3 bff_positioning/visualization/plot_histogram.py /path/to/config.yaml

If I use an example that already exists (like cnn_experiment.yaml), should I just skip the first 2 steps and it will run correctly ?
Thanks a lot for helping!

[gante]

Yeah, the first two steps must be executed in a separate machine to train the model :) After you have a trained model, copy it to the Jetson machine and run the test_model.py step