search

marcelbinz / CENTaUR

18 stars 3 forks source link

Location of model=65B #1

Open gdgdandsz opened 2 months ago

gdgdandsz commented 2 months ago

Hi there, Thank you for this amazing project and I'm currently trying to reproduce your results and adjust some of the prompts. I'm wondering if we have direct access to model=7B,13B,30B and 65B, or if we need to produce them by ourselves?

marcelbinz commented 2 months ago

Hi,

the present code is quite outdated. Nowadays, I would recommend implement everything using Huggingface. We have a tutorial on how to do this with some of the data from the CENTaUR paper here: https://github.com/Zak-Hussain/LLM4BeSci/tree/main/choice

More info in the corresponding paper: https://osf.io/preprints/psyarxiv/f7stn

You should be able to simply plug in the LLaMA models from huggingface: https://huggingface.co/docs/transformers/main/en/model_doc/llama or use some of the newer models instead.

gdgdandsz commented 2 months ago

Thank you for your reply! I will check that.