【大模型】Llama3-8B报错_sentencepiece.SentencePieceProcessor_LoadFromFile(self, arg) TypeError:not a string

环境

• Ubuntu22.04
• python3.11

运行模型

Llama3-8B-Chinese-Chat下载地址：

https://hf-mirror.com/shenzhi-wang/Llama3-8B-Chinese-Chat

跑 Llama3-8B-Chinese-Chat 模型：

from transformers import LlamaTokenizer, AutoModelForCausalLM

model_id = "shenzhi-wang/Llama3-8B-Chinese-Chat"

tokenizer = LlamaTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id, torch_dtype="auto", device_map="auto"
)

messages = [
    {"role": "user", "content": "写一首诗吧"},
]

input_ids = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_tensors="pt"
).to(model.device)

outputs = model.generate(
    input_ids,
    max_new_tokens=8192,
    do_sample=True,
    temperature=0.6,
    top_p=0.9,
)
response = outputs[0][input_ids.shape[-1]:]
print(tokenizer.decode(response, skip_special_tokens=True))

错误信息

File /usr/local/lib/python3.11/dist-packages/sentencepiece/__init__.py:317, in SentencePieceProcessor.LoadFromFile(self, arg)
    315 def LoadFromFile(self, arg):
    316     print("debug: arg is ", arg) 
--> 317     return _sentencepiece.SentencePieceProcessor_LoadFromFile(self, arg)
TypeError: not a string

分析错误

在 Llama、Llama2 中使用 LlamaTokenizer 加载 Tokenizer，但在 Llama3 中使用 AutoTokenizer 加载 Tokenizer

Llama3 is a different tokenizer and should only be initialized with AutoTokenizer

解决方案

修改代码：

from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "shenzhi-wang/Llama3-8B-Chinese-Chat"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id, torch_dtype="auto", device_map="auto"
)

messages = [
    {"role": "user", "content": "写一首诗吧"},
]

input_ids = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_tensors="pt"
).to(model.device)

outputs = model.generate(
    input_ids,
    max_new_tokens=8192,
    do_sample=True,
    temperature=0.6,
    top_p=0.9,
)
response = outputs[0][input_ids.shape[-1]:]
print(tokenizer.decode(response, skip_special_tokens=True))

测试

再次运行模型正常了

参考

• https://github.com/huggingface/transformers/issues/30607