yao8839836 / kg-llm

Exploring large language models for knowledge graph completion
104 stars 8 forks source link

使用lora微调llama结果不对 #8

Closed lv-otfse closed 1 month ago

lv-otfse commented 5 months ago

大佬,您好,我在执行python finetune_wn11.py的代码时损失函数loss很大,训练完保存的权重文件如下,而且adapter_model.bin文件的大小只有1KB,训练的结果不对。我想请教一下您什么原因,是transformers和peft版本的原因吗?我用的是transformers==4.28.0,peft==0.5.0。 屏幕截图 2024-06-08 105246 屏幕截图 2024-06-08 105345 微信图片_20240613103012

hhy-huang commented 3 months ago

I met the same problem, but my loss was not strange. The generated responses includes instruction sentence.