IAI Group level policy for tracking ML training

[IKostric]

Compare tensorboard, W&B, ...

[vinaysetty]

My recommendation:

1. While training use Weights and Biases https://wandb.ai/site to track the hyperparameters and logs
2. Use dataset version control (DVC) https://dvc.org to track datasets used and models trained for the experiments in a GitHub repository and a remote storage like dropbox or even ssh store like on gustav1. Here is the quick start guide https://dvc.org/doc/start/data-and-model-versioning

[kbalog]

Let's try these for TREC CAsT

[kbalog]

Just go with the wandb suggestion for now