Exploring Neptune
Neptune is a Python (and R) package that acts as a metadata store for MLOps. This package supports a lot of features for working with the model-building metadata. We can utilize Neptune for tracking our experiments, not only hyperparameter tuning experiments but also other model-building-related experiments. We can log, visualize, organize, and manage our experiments just by using a single package. Furthermore, it also supports model registry and live monitors our ML jobs.
Installing Neptune is very easy – you can just use pip install neptune-client
or conda install -c conda-forge neptune-client
. Once it has been installed, you need to sign up for an account to get the API token. Neptune is free for an individual plan within the quota limit, but you need to pay if you want to utilize Neptune for commercial team usage. Further information about registering yourself for Neptune can be found on their official website: https://neptune.ai/register.
Using Neptune...