search

facebookresearch / belebele

Repo for the Belebele dataset, a massively multilingual reading comprehension dataset.
Other
312 stars 21 forks source link

Example code #2

Closed geronimi73 closed 8 months ago

geronimi73 commented 1 year ago

would be great if you could share the code you used to evaluate llama2 for example just a few lines. please 😘

lucasbandarkar commented 1 year ago

Hey @geronimi73, please refer to Section 4.2 for the paper, it has all the details of how we evaluated Llama 2. We used internal Meta tooling for it, but we describe the exact prompts we used.

We are working with HF to maybe display some code for it, but that is still TBD