how to fine-tune to improve the result of automatic_mask_generator_example.ipynb

in this https://encord.com/blog/learn-how-to-fine-tune-the-segment-anything-model-sam/ it say
[Finally, we can generate the masks. Note that here we are in single mask generation mode (in contrast to the 3 masks that are normally output).]
with torch.no_grad():
sparse_embeddings, dense_embeddings = sam_model.prompt_encoder(
points=None,
boxes=box_torch,
masks=None,
)
Does it mean that the parameter masks=False must be specified during fine-tune?
and parameter boxes must be specified ?

Start with automatic_mask_generator_example.ipynb and compare its mask-generation flow with the fine-tuning example in the linked Encord article. Determine how the notebook's prompt_encoder parameters relate to the question about masks and boxes, then document the correct fine-tuning guidance. The issue is done when the parameter requirements and expected usage are clearly explained.

Written by the indexing model from the issue text.