Regarding the Dataset

#2
by abdullahalzubaer - opened

Hi, I was wondering, for the instruction fine-tuning, which dataset was used.

Thank you.

Hi,
The model has been trained on yizhongw/self_instruct translated to German.
There has been no further processing on the translated data, so there is potential for improvement. But model quality does not matter for us at the moment, we are still in the process of evaluating different train/inference strategies.
Cheers
Andre

Hello,

Thank you for the reply. I have already tested your adapter, and it was performing reasonably good already. I will close the issue since my question has been answered :)

abdullahalzubaer changed discussion status to closed

Sign up or log in to comment