The fine-tuned model (.pth file) is very large

InternLM / InternLM-XComposer

InternLM-XComposer2 is a groundbreaking vision-language large model (VLLM) excelling in free-form text-image composition and comprehension.

1.91k stars 120 forks source link

Closed nzomi closed 2 weeks ago

nzomi commented 2 weeks ago

The fine-tuned file in the finetune/output is very large, around 2 times(35G) than the base pth(17G), dose anyone come across a same question?

nzomi commented 2 weeks ago

already done