推理报错

[dongguanting]

训练后加载模型推理时报错，请问这是什么原因

Traceback (most recent call last):
File "/cpfs01/shared/KnowPAT/inference.py", line 68, in
model = LlamaForCausalLM.from_pretrained(
File "/opt/conda/envs/llama_factory/lib/python3.10/site-packages/transformers/modeling_utils.py", line 3787, in from_pretrained
model.load_adapter(
File "/opt/conda/envs/llama_factory/lib/python3.10/site-packages/transformers/integrations/peft.py", line 193, in load_adapter
adapter_state_dict = load_peft_weights(peft_model_id, token=token, **adapter_kwargs)
File "/opt/conda/envs/llama_factory/lib/python3.10/site-packages/peft/utils/save_and_load.py", line 244, in load_peft_weights
has_remote_safetensors_file = file_exists(
File "/opt/conda/envs/llama_factory/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 110, in _inner_fn
validate_repo_id(arg_value)
File "/opt/conda/envs/llama_factory/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 158, in validate_repo_id
raise HFValidationError(
huggingface_hub.utils._validators.HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/KnowPAT/checkpoint/nq_5k/checkpoint-1800/'. Use repo_type argument if needed.

加载的checkpoint如下：

[hzrlsd]

这是来自QQ邮箱的自动回复邮件。   您好，现在无法亲自回复您的邮件。我会尽快给您回复。

[Zhang-Each]

inference部分的代码默认加载的是训练完成后的模型文件中的adapter_model.bin，如果要加载checkpoint，可以尝试把pytorch_model.bin改名为adapter_model.bin

[dongguanting]

请问inference.py中，save path和model path分别填入什么呢？

我猜测：
save path填入checkpoint文件夹的path
model path直接填入原始llama的path吗？

[Zhang-Each]

你好，是的