Fill-Mask
Transformers
PyTorch
modernbert

safetensors

#15
by bwallima - opened

Why was Safetensors removed, and why hasn't the current pull request been accepted?

Center for Language and Speech Processing @ JHU org

Hey @bwallima ! When I first was putting this up and merged it, it didn't save the tensors right. Something about the conversion at the time made it change the model (and be much worse).

If you have time to double check that the safetensors output is equivalent to the regular output, would be happy to merge this! Just haven't had time to do it.

Sign up or log in to comment