search

fe1ixxu / ALMA

State-of-the-art LLM-based translation models.
MIT License
352 stars 26 forks source link

Question on the model checkpoint #1

Closed gpengzhi closed 9 months ago

gpengzhi commented 9 months ago

Great work!

Is there any plan to release the model checkpoint only with monolingual data finetuning?

fe1ixxu commented 9 months ago

Thanks for the interest! haoranxu/ALMA-7B-Pretrain and haoranxu/ALMA-13B-Pretrain are the models which are only fine-tuned on monolingual data. To clarify, haoranxu/ALMA-7B-Pretrain-LoRA and haoranxu/ALMA-13B-Pretrain-LoRA are just LoRA for them in the MT task~