Instructions to use HuggingFaceM4/idefics2-8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HuggingFaceM4/idefics2-8b with Transformers:

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

pipe = pipeline("image-text-to-text", model="HuggingFaceM4/idefics2-8b")

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("HuggingFaceM4/idefics2-8b")
model = AutoModelForMultimodalLM.from_pretrained("HuggingFaceM4/idefics2-8b")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use HuggingFaceM4/idefics2-8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HuggingFaceM4/idefics2-8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HuggingFaceM4/idefics2-8b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/HuggingFaceM4/idefics2-8b

SGLang

How to use HuggingFaceM4/idefics2-8b 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 "HuggingFaceM4/idefics2-8b" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HuggingFaceM4/idefics2-8b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "HuggingFaceM4/idefics2-8b" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HuggingFaceM4/idefics2-8b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use HuggingFaceM4/idefics2-8b with Docker Model Runner:
```
docker model run hf.co/HuggingFaceM4/idefics2-8b
```

Fine-tuned idefics for inference?

#37

by baldesco - opened May 1, 2024

Discussion

baldesco

May 1, 2024

I followed the fine-tuning tutorial and pushed to hugging face.

How can I use this fine-tuned version for inference?

VictorSanh

May 2, 2024

Hi @baldesco ,
assuming you are following the tutorial, then you saved some lora parameters.
you can load the model trained with its lora parameters simply by doing a AutoModel.from_pretrained(PATH_TO_YOUR_MODEL_ON_THE_HUB)

baldesco

May 2, 2024

Thank you @VictorSanh for the answer.

Yes, I followed the tutorial so I am using QLORA.

In the sample code for using idefics2 for inference, this is how the model is loaded:

processor = AutoProcessor.from_pretrained("HuggingFaceM4/idefics2-8b")
model = AutoModelForVision2Seq.from_pretrained(
    "HuggingFaceM4/idefics2-8b",
).to(DEVICE)

So I have 2 questions:

For the processor, should I leave the original idefics model, or should I also point to my weights?
For the model you mention AutoModel.from_pretrained, but the snippet above uses AutoModelForVision2Seq.from_pretrained. Are these two the same, or when should I use each?

Thank you

VictorSanh

May 2, 2024

1/ either way. that is equivalent
2/ AutoModelForVision2Seq is indeed safer, if you want to be really safe (i.e. avoid any mismatch in the auto-mapping), I would even recommend Idefics2ForConditionalGeneration.from_pretrained

VictorSanh changed discussion status to closed May 3, 2024

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment