按照文档导出Qwen2-0.5B-Instruct mnn报错

[fengge123]

平台(如果交叉编译请再附上交叉编译目标平台):

Platform(Include target platform as well if cross-compiling):

wsl ubuntu2204

Github版本:

Github Version:

commit 2b899c1 (HEAD -> master, origin/master, origin/HEAD)

直接下载ZIP包请提供下载日期以及压缩包注释里的git版本(可通过7z l zip包路径命令并在输出信息中搜索Comment 获得，形如Comment = bc80b11110cd440aacdabbf59658d630527a7f2b)。 git clone请提供 git commit 第一行的commit id

Provide date (or better yet, git revision from the comment section of the zip. Obtainable using 7z l PATH/TO/ZIP and search for Comment in the output) if downloading source as zip,otherwise provide the first commit id from the output of git commit

编译方式:

Compiling Method

cd MNN/pymnn/pip_package
python3 build_deps.py llm
python3 setup.py install --deps llm --prefix=/mnt/d/workspace/MNN/install/python/

请在这里粘贴cmake参数或使用的cmake脚本路径以及完整输出
Paste cmake arguments or path of the build script used here as well as the full log of the cmake proess here or pastebin

编译日志:

Build Log:

粘贴在这里
Paste log here or pastebin

问题：

我参考下面文档，测试export 工具，
https://mnn-docs.readthedocs.io/en/latest/transformers/llm.html
我的命令是：
git clone https://www.modelscope.cn/qwen/Qwen2-0.5B-Instruct.git
存放到/mnt/d/workspace/llm-models/Qwen2-0.5B-Instruct
cd ./transformers/llm/export
python3 llmexport.py --path /mnt/d/workspace/llm-models/Qwen2-0.5B-Instruct/ --export mnn

报错：
💥 Failed load pretrained model
Traceback (most recent call last):
File "/mnt/d/workspace/MNN/transformers/llm/export/llmexport.py", line 2291, in load_pretrained
self.model = AutoModelForCausalLM.from_pretrained(model_path, torch_dtype='auto', trust_remote_code=True).eval()
File "/home/fg/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 564, in from_pretrained
return model_class.from_pretrained(
File "/home/fg/.local/lib/python3.10/site-packages/transformers/modeling_utils.py", line 4010, in from_pretrained
with safe_open(resolved_archive_file, framework="pt") as f:
safetensors_rust.SafetensorError: Error while deserializing header: HeaderTooLarge

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/mnt/d/workspace/MNN/transformers/llm/export/llmexport.py", line 44, in wrapper
result = func(*args, **kwargs)
File "/mnt/d/workspace/MNN/transformers/llm/export/llmexport.py", line 2306, in load_model
self.load_pretrained(model_path)
File "/mnt/d/workspace/MNN/transformers/llm/export/llmexport.py", line 2293, in load_pretrained
self.model = AutoModel.from_pretrained(model_path, torch_dtype='auto', trust_remote_code=True).eval()
File "/home/fg/.local/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py", line 564, in from_pretrained
return model_class.from_pretrained(
File "/home/fg/.local/lib/python3.10/site-packages/transformers/modeling_utils.py", line 4010, in from_pretrained
with safe_open(resolved_archive_file, framework="pt") as f:
safetensors_rust.SafetensorError: Error while deserializing header: HeaderTooLarge

补充:wsl ubuntu分配的是32G内存

[fengge123]

模型没有下完整 ，问题解决

[jxt1234]

你应该没有安装 git lfs ，建议简单起见 pip install modelscope ，用modelscope 安装