Facico / Chinese-Vicuna

Chinese-Vicuna: A Chinese Instruction-following LLaMA-based Model —— 一个中文低资源的llama+lora方案,结构参考alpaca
https://github.com/Facico/Chinese-Vicuna
Apache License 2.0
4.14k stars 421 forks source link

无法成功执行generate.sh以及interaction.sh #55

Closed alisyzhu closed 1 year ago

alisyzhu commented 1 year ago

完成fine-tune后,按照readme里配置好脚本参数后,执行predict报错,烦请帮忙看看是哪里配置没有修改吗? 1、sh generate.sh

image

debug后发现,在GPU加载完LLAMA模型后,执行

image

后model=None了。。 2、sh interaction.sh

image image
alisyzhu commented 1 year ago

大概知道是什么问题了 ~

Facico commented 1 year ago

抱歉这么晚回复,这个问题是peft版本的问题,可以参考这个issue