Instructions to use openbmb/MiniCPM-Llama3-V-2_5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use openbmb/MiniCPM-Llama3-V-2_5 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="openbmb/MiniCPM-Llama3-V-2_5", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("openbmb/MiniCPM-Llama3-V-2_5", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use openbmb/MiniCPM-Llama3-V-2_5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "openbmb/MiniCPM-Llama3-V-2_5"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "openbmb/MiniCPM-Llama3-V-2_5",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/openbmb/MiniCPM-Llama3-V-2_5

SGLang

How to use openbmb/MiniCPM-Llama3-V-2_5 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "openbmb/MiniCPM-Llama3-V-2_5" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "openbmb/MiniCPM-Llama3-V-2_5",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "openbmb/MiniCPM-Llama3-V-2_5" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "openbmb/MiniCPM-Llama3-V-2_5",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use openbmb/MiniCPM-Llama3-V-2_5 with Docker Model Runner:
```
docker model run hf.co/openbmb/MiniCPM-Llama3-V-2_5
```

突然加载不了tokenizer了，是不是升级导致的版本不兼容

#67

by Jiiao - opened Jul 30, 2024

Discussion

Jiiao

Jul 30, 2024

AttributeError Traceback (most recent call last)
测试.ipynb Cell 19 line 4
1 # vpm_resampler_embedtokens_weight = torch.load(f"{model_dir}/vpm_resampler_embedtokens.pt")
2
3 # msg = model.load_state_dict(vpm_resampler_embedtokens_weight, strict=False)
----> 4 tokenizer = AutoTokenizer.from_pretrained(model_dir, trust_remote_code=True)

File /usr/local/lib/python3.9/dist-packages/transformers/models/auto/tokenization_auto.py:877, in AutoTokenizer.from_pretrained(cls, pretrained_model_name_or_path, *inputs, **kwargs)
875 else:
876 class_ref = tokenizer_auto_map[0]
--> 877 tokenizer_class = get_class_from_dynamic_module(class_ref, pretrained_model_name_or_path, **kwargs)
878 _ = kwargs.pop("code_revision", None)
879 if os.path.isdir(pretrained_model_name_or_path):

File /usr/local/lib/python3.9/dist-packages/transformers/dynamic_module_utils.py:514, in get_class_from_dynamic_module(class_reference, pretrained_model_name_or_path, cache_dir, force_download, resume_download, proxies, token, revision, local_files_only, repo_type, code_revision, **kwargs)
501 # And lastly we get the class inside our newly created module
502 final_module = get_cached_module_file(
503 repo_id,
504 module_file + ".py",
ref='/usr/local/lib/python3.9/dist-packages/transformers/dynamic_module_utils.py:0'>0;32m (...)
512 repo_type=repo_type,
513 )
--> 514 return get_class_in_module(class_name, final_module)

File /usr/local/lib/python3.9/dist-packages/transformers/dynamic_module_utils.py:213, in get_class_in_module(class_name, module_path)
211 # reload in both cases
212 module_spec.loader.exec_module(module)
--> 213 return getattr(module, class_name)

AttributeError: module 'transformers_modules.openbmb.MiniCPM-Llama3-V-2_5.287e3f85192a7c4acf2564fc6bda0637439a9d78.modeling_minicpmv' has no attribute 'PreTrainedTokenizerFastWrapper'

finalf0

Jul 30, 2024

•

edited Jul 30, 2024

请问model_dir 是有小数点吗？MiniCPM-Llama3-V-2_5.287e3f85192a7c4acf2564fc6bda0637439a9d78 这样？由于huggingface的机制问题，model_dir中带小数点会导致dynamic import错误，请把小数点改成其他符号

Jiiao

Jul 30, 2024

请问model_dir 是有小数点吗？MiniCPM-Llama3-V-2_5.287e3f85192a7c4acf2564fc6bda0637439a9d78 这样？由于huggingface的机制问题，model_dir中带小数点会导致dynamic import错误，请把小数点改成其他符号

hi，谢谢您的回复！
model_dir中没有小数点，我加载的代码如下：

       self.model = AutoPeftModelForCausalLM.from_pretrained(model_dir,
                                                            device_map='cuda:0',
                                                            trust_remote_code=True,
                                                            torch_dtype=torch.float16
                                                        ).eval()

        vpm_resampler_embedtokens_weight = torch.load(f"{model_dir}/vpm_resampler_embedtokens.pt")

        self.msg = self.model.load_state_dict(vpm_resampler_embedtokens_weight, strict=False)
        self.tokenizer = AutoTokenizer.from_pretrained(model_dir, trust_remote_code=True)

在7天前的commit（287e3f8）之后，代码执行会报以上错误，我目前理解的过程是，因为用PEFT加载模型基本上必须要 trust_remote_code=True，每次会拉最新的代码，也就是transformers_modules.openbmb.MiniCPM-Llama3-V-2_5.287e3f85192a7c4acf2564fc6bda0637439a9d78.modeling_minicpmv；最近这个版本的提交，tokenizer的定义中少了PreTrainedTokenizerFastWrapper，导致加载失败，我自己修改需要把整个仓库clone下来自己修复，后续跟进不了最新版本的代码，不知道是否你们能够修复呢，或者如何避免呢