Add a new tensor to existing safetensors file

[kentwait]

What is the recommended way to add a new tensor to an existing safetensors file?

For example, I load the existing safetensors file into dictionary d using the torch API. Then I add a new tensor with the key "test" into the dictionary. Can I save the modified dictionary of tensors using the same filename? Safetensors is memory-mapped so I assume it's not safe but I'm asking for clarification or suggestion as to what is the best way to do this kind of update.

from safetensors.torch import load_file, save_file
import torch

# Load
file_path = "embedding.safetensors"
d = load_file(file_path)

# Add entry
d['test'] = torch.tensor([0.0, 0.1, 0.2])

# Save to same filename ok?
save_file(d, file_path)