Instructions to use meta-llama/Llama-3.2-11B-Vision with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use meta-llama/Llama-3.2-11B-Vision with Transformers:

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

pipe = pipeline("image-text-to-text", model="meta-llama/Llama-3.2-11B-Vision")

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("meta-llama/Llama-3.2-11B-Vision")
model = AutoModelForMultimodalLM.from_pretrained("meta-llama/Llama-3.2-11B-Vision", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use meta-llama/Llama-3.2-11B-Vision with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "meta-llama/Llama-3.2-11B-Vision"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "meta-llama/Llama-3.2-11B-Vision",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/meta-llama/Llama-3.2-11B-Vision

SGLang

How to use meta-llama/Llama-3.2-11B-Vision 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 "meta-llama/Llama-3.2-11B-Vision" \
    --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": "meta-llama/Llama-3.2-11B-Vision",
		"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 "meta-llama/Llama-3.2-11B-Vision" \
        --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": "meta-llama/Llama-3.2-11B-Vision",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use meta-llama/Llama-3.2-11B-Vision with Docker Model Runner:
```
docker model run hf.co/meta-llama/Llama-3.2-11B-Vision
```

Re-download the llama 11B HF format but suddenly hit error in loading model

#16

by hxgy610 - opened Sep 25, 2024

Discussion

hxgy610

Sep 25, 2024

Error message is as below.

File "/opt/conda/lib/python3.10/site-packages/transformers/modeling_utils.py", line 4507, in _load_pretrained_model
    raise RuntimeError(f"Error(s) in loading state_dict for {model.__class__.__name__}:\n\t{error_msg}")
RuntimeError: Error(s) in loading state_dict for MllamaForConditionalGeneration:
        size mismatch for vision_model.gated_positional_embedding.embedding: copying a param with shape torch.Size([1025, 1280]) from checkpoint, the shape in current model is torch.Size([1601, 1280]).
        size mismatch for vision_model.gated_positional_embedding.tile_embedding.weight: copying a param with shape torch.Size([9, 5248000]) from checkpoint, the shape in current model is torch.Size([9, 8197120]).

I used the latest transformers transformers-20240924-4.45.0.dev0-py3-none-any.whl

pcuenq

Sep 25, 2024

Should be fixed now, there was a change in the configuration (448 to 560 image size) that required some additional adjustments in the tile embeddings. You have to download the weights again (sorry), but it should work with the same wheel.

osanseviero changed discussion status to closed Sep 25, 2024

hxgy610

Sep 26, 2024

•

edited Sep 26, 2024

Hi @pcuenq I confirmed that model loading is working again, but do you know any other potential factor impacting memory usage? The same finetuning code now goes OOM on the same hardware. I am using transformer 4.45.0

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