example-wandb-trainer

This repository is an example of training repository which integrate with WandB for login. This repo use dataset which is iris.csv also being used from this repository https://github.com/alphinside/example-dvc-dataset Please install poetry to install all the dependencies

How to use :

1. Install dependencies

poetry install

2. Login to wandb

poetry run wandb login

3. Run the script, notes that the --dataset-path default is refer to another repository dir mentioned above in writer local system

poetry run python example_wandb_trainer/main.py --help
Usage: main.py [OPTIONS]

Options:
  --dataset-path TEXT             Path to iris csv dataset  [default:
                                  ../example-dvc-dataset/dataset/iris.csv]
  --wandb-entity TEXT             WandB user profile  [required]
  --experiment-type [svc|decision_tree]
                                  Experiment model type  [required]

E.g.

poetry run python example_wandb_trainer/main.py --wandb-entity alvin-prayuda --experiment-type svc

Example output:

wandb: Synced svc: https://wandb.ai/alvin-prayuda/experiment-tracking-example/runs/173aitq9
wandb: Synced 5 W&B file(s), 5 media file(s), 5 artifact file(s) and 1 other file(s)
wandb: Find logs at: ./wandb/run-20220322_165209-173aitq9/logs

4. The dumped model will be available at ./tmp directory on your local working directory