[DOC] Fine-tuning options are almost undocumented #1027
Description
Location of the issue
See e.g. here https://docs.metatensor.org/metatrain/latest/architectures/generated/pet.html#trainer-hyperparameters
Description of the issue
They're basically undocumented
Suggested fix or improvement (optional)
We should provide a description of each possible fine-tuning parameter
Additional context
No response