what is the prompt used for instruction tuning, and why the model is pre-trained on refineweb but also instruction-tuned with it?

#30
by zerolyn - opened

can i have the prompts for instruction tuning? I'm also confused about why the model is pre-trained with refineweb but also instruction-tuned with its subset?

Hi, as requested by @zerolyn , would be very useful to have the exact prompt used for fine-tuning to make the generation more reliable and robust. Thanks a lot for your contribution to the open-source!

Can we please get the prompt used for fine-tuning.

Sign up or log in to comment