Lora微调大模型找不到Template的问题

问题提出

使用Lora进行微调大模型时，终端报如下错误：

Traceback (most recent call last):
  File "/usr/local/bin/llamafactory-cli", line 33, in <module>
    sys.exit(load_entry_point('llamafactory==0.7.2.dev0', 'console_scripts', 'llamafactory-cli')())
  File "/usr/local/lib/python3.10/site-packages/llamafactory-0.7.2.dev0-py3.10.egg/llamafactory/cli.py", line 65, in main
    run_exp()
  File "/usr/local/lib/python3.10/site-packages/llamafactory-0.7.2.dev0-py3.10.egg/llamafactory/train/tuner.py", line 34, in run_exp
    run_sft(model_args, data_args, training_args, finetuning_args, generating_args, callbacks)
  File "/usr/local/lib/python3.10/site-packages/llamafactory-0.7.2.dev0-py3.10.egg/llamafactory/train/sft/workflow.py", line 33, in run_sft
    dataset = get_dataset(model_args, data_args, training_args, stage="sft", **tokenizer_module)
  File "/usr/local/lib/python3.10/site-packages/llamafactory-0.7.2.dev0-py3.10.egg/llamafactory/data/loader.py", line 123, in get_dataset
    template = get_template_and_fix_tokenizer(tokenizer, data_args.template)
  File "/usr/local/lib/python3.10/site-packages/llamafactory-0.7.2.dev0-py3.10.egg/llamafactory/data/template.py", line 352, in get_template_and_fix_tokenizer
    raise ValueError("Template {} does not exist.".format(name))

解决方案：

终端输入：

vim /usr/local/lib/python3.10/site-packages/llamafactory-0.7.2.dev0-py3.10.egg/llamafactory/data/template.py

在文件最后添加模板即可：

_register_template(
    name="model_name",
    format_user=StringFormatter(slots=["<用户>{{content}}<AI>"]),
    format_system=StringFormatter(slots=[{"bos_token"}, "{{content}}"]),
    force_system=True,
)