How to apply this methods to a new LMs without fine-tune, like a bert?

kmeng01 / rome

Locating and editing factual associations in GPT (NeurIPS 2022)

https://rome.baulab.info

MIT License

550 stars 110 forks source link

How to apply this methods to a new LMs without fine-tune, like a bert? #28

Open sev777 opened 1 year ago

sev777 commented 1 year ago

Hi, In your paper, does the model edited in ROME or MEND methods must be fine-tuned? How to apply the ROME to the model without fine-tuning ? For example, the bert model is not just the GPTs series.

Thanks.

huu4ontocord commented 1 year ago

Same question. How to apply to t5. I see that the MEND algorithm can be applied to t5, but not ROME.

Wanglvcheng commented 1 week ago

我也想问这个问题