Neural network interpretability ; Neural network interpretability AND Princeton
Common descendants
1 Document
Sanjeev Arora sur Twitter : "new `skills' induced by LLM fine-tuning can be localized in tiny fraction of the model."
Tags:
2023-07-07 About