Difficulty to follow the Action Recognition Net tutorial using TAO

[guruh20001]

ENVIRONMENT

Computer spec:
Proc: AMD R9-9900X
Memory: 32 GB
GPU: 5060 Ti

OS: Ubuntu 24.04.2 LTS

Python: 3.13.5
docker-ce: 28.3.0
docker-API: 1.51
nvidia-container-toolkit: 1.18.0-rc.1
nvidia-docker2: 2.14.0-1
nvidia-driver: 570.153.02

===================================================

I have a difficulty to follow the Action Recognition Net tutorial using TAO in “tao_tutorials/notebooks/tao_launcher_starter_kit/action_recognition_net/actionrecognitionnet.ipynb”. When i use the command…

print(‘Encrypted checkpoints:’)
print(‘---------------------’)
!ls -ltrh $HOST_RESULTS_DIR/rgb_3d_ptm/train

…i got the error…

Encrypted checkpoints:
< ---------------------
ls: cannot access ‘/rgb_3d_ptm/train’: No such file or directory

I didn’t receive any error on previous commands. Where did i do it wrong?

Thanks.

[Morganh]

@guruh20001

!ls -ltrh $HOST_RESULTS_DIR/rgb_3d_ptm/train

@guruh20001

ls: cannot access ‘/rgb_3d_ptm/train’: No such file or directory

Please double check if you set $HOST_RESULTS_DIR.

[guruh20001]

Thanks Morghans for the input. I already double checked and it’s seems already “set”. I decided to restart from the beginning, and this time i got the error when i run the command…

!cat $HOST_SPECS_DIR/experiment_rgb_3d_finetune.yaml

…produce the error…

cat: /home/guruh/specs/experiment_rgb_3d_finetune.yaml: No such file or directory

I suspect this error dan previous error is just maybe i’m doing wrong to set the working directory. The yaml file error is in directory “/home/guruh/specs/…” buy my local project is “LOCAL_PROJECT_DIR=/home/guruh/tao-experiments/”. Am i read the code wrong?

I attach the screenshot of the Jupyter notebook for further reference.

Thank you.

Screenshot 2025-07-06 174905.pdf (465.9 KB)

[Morganh]

Please check the ~/tao_mounts.json. What is the content?

[guruh20001]

Hi Morganh, below is the content of tao_mounts.json

[Morganh]

Suggest you to set the same source and destination.
And only set only one.
i.e.,

"source": "/home/guruh",
"destination": "/home/guruh"

Then, all the files under /home/guruh will be mapped into the docker.

[yingliu]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks