peft set_adapter报错

[guxinyu1225]

Describe the bug/ 问题描述 (Mandatory / 必填)
使用peft加载多个lora adapter后，使用set_adapter尝试激活其中一个lora adapter后报错AttributeError: 'LlamaForCausalLM' object has no attribute 'cells'

• Hardware Environment(Ascend/GPU/CPU) / 硬件环境:

Please delete the backend not involved / 请删除不涉及的后端:
/device ascend/GPU/CPU/kirin/等其他芯片

• Software Environment / 软件环境 (Mandatory / 必填):
  -- MindSpore version (e.g., 1.7.0.Bxxx) : 2.3.0
  -- Python version (e.g., Python 3.7.5) : 3.9.10
  -- OS platform and distribution (e.g., Linux Ubuntu 16.04):Ubuntu 18.04

• Excute Mode / 执行模式 (Mandatory / 必填)(PyNative/Graph):

/mode pynative

To Reproduce / 重现步骤 (Mandatory / 必填)
在加载tloen/alpaca-lora-7b以及22h/cabrita-lora-v0-1前，我用

def torch_to_mindspore(ckpt_path, save_path):
state_dict = torch.load(ckpt_path, map_location="cpu")
ms_ckpt = []
for k, v in state_dict.items():
if 'wq' in k:
k = k.replace('wq', 'w_q')
v = v.transpose(0, 1)
if 'wk' in k:
k = k.replace('wk', 'w_k')
v = v.transpose(0, 1)
if 'wv' in k:
k = k.replace('wv', 'w_v')
v = v.transpose(0, 1)
if 'wo' in k:
k = k.replace('wo', 'w_o')
v = v.transpose(0, 1)
if 'w1' in k:
k = k.replace('w1', 'w_1')
v = v.transpose(0, 1)
if 'w2' in k:
k = k.replace('w2', 'w_2')
v = v.transpose(0, 1)
if 'w3' in k:
k = k.replace('w3', 'w_3')
v = v.transpose(0, 1)
if 'output' in k:
v = v.transpose(0, 1)
if 'rope' in k:
continue
ms_ckpt.append({'name': k, 'data': mindspore.Tensor(v.numpy())})
mindspore.save_checkpoint(ms_ckpt, save_path)

将adapter_model.bin转为adapter_model.ckpt

from mindnlp.transformers import LlamaForCausalLM, LlamaTokenizer
from mindnlp.peft import PeftModel, LoraConfig
model_name = "baffo32/decapoda-research-llama-7B-hf"
tokenizer = LlamaTokenizer.from_pretrained(model_name)
model = LlamaForCausalLM.from_pretrained(model_name)
model = PeftModel.from_pretrained(model, "tloen/alpaca-lora-7b", adapter_name="eng_alpaca")
peft_config = LoraConfig.from_pretrained("22h/cabrita-lora-v0-1")
model.add_adapter(adapter_name="portuguese_alpaca",peft_config=peft_config)
model.load_adapter("./cabrita-lora-v0-1", adapter_name="portuguese_alpaca")
model.set_adapter("eng_alpaca")

Expected behavior / 预期结果 (Mandatory / 必填)
eng_alpaca adapter激活

Screenshots/ 日志 / 截图 (Mandatory / 必填)

Additional context / 备注 (Optional / 选填)
Add any other context about the problem here.