Azurro

/

llama-2-7b-qlora-polish

like 1

Follow

Azurro 11

Probably it is worse, because it was trained on smaller (instructions) dataset and it was qlora vs full finetuning.