microsoft / mttl

Building modular LMs with parameter-efficient fine-tuning.
MIT License
62 stars 3 forks source link

Alpaca #23

Closed pclucas14 closed 2 months ago

pclucas14 commented 1 year ago

Adding the alpaca dataset to the repo.

Right now, valid set is a fixed random split from the full training data. There is no test set (we are reusing the dev split)