基于qwen模型使用coba训练后，权重合并错误

[ChaohuaZhang]

Traceback (most recent call last):
File "/data/demo/MFTCoder/mftcoder_accelerate/src/pefts/merge_base_and_lora_to_hf.py", line 82, in
model_to_merge = PeftModel.from_pretrained(base_model, lora_adapter)
File "/root/miniconda3/envs/mft/lib/python3.9/site-packages/peft/peft_model.py", line 355, in from_pretrained
model = MODEL_TYPE_TO_PEFT_MODEL_MAPPING[config.task_type](model, config, adapter_name)
File "/root/miniconda3/envs/mft/lib/python3.9/site-packages/peft/peft_model.py", line 1094, in init
super().init(model, peft_config, adapter_name)
File "/root/miniconda3/envs/mft/lib/python3.9/site-packages/peft/peft_model.py", line 129, in init
self.base_model = cls(model, {adapter_name: peft_config}, adapter_name)
File "/root/miniconda3/envs/mft/lib/python3.9/site-packages/peft/tuners/lora/model.py", line 136, in init
super().init(model, config, adapter_name)
File "/root/miniconda3/envs/mft/lib/python3.9/site-packages/peft/tuners/tuners_utils.py", line 148, in init
self.inject_adapter(self.model, adapter_name)
File "/root/miniconda3/envs/mft/lib/python3.9/site-packages/peft/tuners/tuners_utils.py", line 328, in inject_adapter
raise ValueError(
ValueError: Target modules {'c_proj', 'w1', 'c_attn', 'w2'} not found in the base model. Please check the target modules and try again.

[GoneZ5]

This issue may not be related to which MFT loss was used. It's possible that the problem stems from an incorrect setting of the model_type (qwen or qwen2). Could you please confirm whether the model_type used during training and merging is consistent?

It's important to note that the 'qwen' model_type cannot be used to load models based on Qwen2 or later versions.