training on base 1.0This LoRA has overfitting to "From above".v1-fix
修复了tlora的bug
nnmbpx_v1_epoch22.safetensors