It is highly and well integrated with libraries I am using like PyTorch Lightning. Many times I have set that up because it was very easy, and after a while, it has happened that I lost part of my results, and W&B helped me to easily recover them through the logs, which without it, I probably wouldn't have. The next and very important feature to me is that I can use many different machines and servers at the same time and without being worried about gathering all results together, then using tools like TensorBoard, having them online without any effort(most of the time saves me when I am using supercomputers). It is part of my daily tools, and when I am teaching students machine learning, in very early sessions after teaching them visualization, I will have them use W&B to repeat whatever they have learned so far. I have never had any problem with W&B, but I have heard from one of my friends, whom I recommended he use W&B, that customer support is very fast and experienced.

I like the flexibility to log custom parameters, and the colorful comparisons between models in for instance confusion matrices.

Everything works almost out of the box and it has a "nice look" once logged in