Simple starts Question for newbie

[vamoruso]

Hello i'm trying to start a train experiment on windows 11
using python 3.10
when executing the following command
ludwig train -c train_conf.yaml --dataset final_train.csv --output_directory . --data_format csv --experiment_name train --verbose
nothing happen.
I only see in console this message :
PyTorch version 2.1.2 available.

This is the content of yaml file :
input_features:

• name: Title
  type: text
• name: Description
  type: text

output_features:

• name: ClassIndex
  type: category

trainer:
# learning_rate: 0.0001
# decay: True
# decay_rate: 0.9
# decay_steps: 30000
# staircase: True
epochs: 20

Version of ludwig i 0.9.0 dev.
Thank you in advance.
Vince

[arnavgarg1]

Hi @vamoruso!

Are you able to try this command instead?

ludwig train -c train_conf.yaml --dataset final_train.csv --output_directory . --data_format csv --experiment_name train --logging_level info

We don't support the verbose argument; instead, we call it logging_level. There are 5 options (but INFO is the one that's recommend if you're not actively developing/contributing to Ludwig): "critical", "error", "warning", "info", "debug".

Let me know if that resolves your issue!

[arnavgarg1]

I see @vamoruso. I tried this on mac and Linux and I can see that it logs to stdout. Unfortunately, I don't have a windows machine so will need to check with the other maintainers and see if anyone can reproduce and debug the issue.

[vamoruso]

Thanks a lot, let me know.

[vamoruso]

I've tried on a friend's MacOS and it works. @arnavgarg1 is there a suggested Linux distro for Ludwig?.
Thank you for job and support.
Ciao
Vincenzo

[arnavgarg1]

Hi @vamoruso! Glad to know you were able to confirm that it works on a Mac.

Ludwig actually ships with Docker images - would that work for you? I believe it uses the official Torch and Ray docker images, both of which use Ubuntu.

[arnavgarg1]

You can check out our various images here: https://hub.docker.com/u/ludwigai

Here's a description of the packages: https://ludwig.ai/latest/getting_started/docker/