Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Is it possible to do further training on the weights they release?


Yes, and there are a sea of finetunes. See: https://huggingface.co/models?sort=modified&search=Ggml

QLORA is the most cost effective method so far. Some people also do finetuning on Google TPUs




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: