[XPU] add xpu support for llama sft#9152

ZHUI merged 1 commit into

PaddlePaddle:developPaddlePaddle/PaddleNLP:developfrom

tizhou86:xpu_lineartizhou86/PaddleNLP:xpu_linearCopy head branch name to clipboard