Refactor UFM methods and implement zeroth order learning in `unadapt.py`

mukobi / Unadaptable-Foundation-Models

MIT License

3 stars 0 forks source link

Closed Rohan138 closed 6 months ago

Rohan138 commented 6 months ago

Refactor unadapt methods to work with the HuggingFace models from #2
Implement zeroth-order learning and benchmark time for zeroth order and gradient methods
Possibly improve memory/speed for both inference and hypergradient calc

Rohan138 commented 6 months ago

Close duplicate

mukobi commented 6 months ago

Dupes #26