peft微调Bloom时报错

报错如下:

peft报错
RuntimeError: self and mat2 must have the same dtype

明确禁用下8bit训练即可

model = AutoModelForCausalLM.from_pretrained(
base_model,
load_in_8bit=False,
torch_dtype=torch.float16,
device_map=device_map,
)
model.is_loaded_in_8bit = False #明确一下，不使用8bit进行训练，因为huggingface的8bit训练不支持
model = prepare_model_for_int8_training(model)