Neptune stores all the runs’ results that later can be gathered into a single dataframe. It is extremely helpful to have all the experiments in one, accessible place. This, combined with run tagging, creates a powerful combination for results analysis. I recommend neptune for ML projects. It is very beginner friendly and easy to introduce in any ML framework. I’ve successfully used it with keras, pytorch, and optuna. I think it is a great tool that is a staple when it comes to experiment management.

the ability to have anything related to a model run in the same spot, with easy to use configuration and modularity.

You can quickly set up your Neptune logging using the flexible python API. The team is responsive and is actively working to improve the UX.