LongLoRA icon indicating copy to clipboard operation
LongLoRA copied to clipboard

Published 20 hours ago • verified publisher icon dvlab-research

Is LongAlpaca model fine-tuned from llama-2 or the Alpaca model?

Open Mooler0410 opened this issue 1 year ago • 0 comments

After carefully reading your huggingface repo, by my understanding, the LongAlpaca model is directly tuned from LLama-2 within 1000 steps on the LongAlpaca dataset. Is that correct?

Mooler0410 avatar Dec 08 '23 04:12 Mooler0410