Instructions to use jadechoghari/Ferret-UI-Llama8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jadechoghari/Ferret-UI-Llama8b with Transformers:

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

pipe = pipeline("image-text-to-text", model="jadechoghari/Ferret-UI-Llama8b", 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 AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("jadechoghari/Ferret-UI-Llama8b", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use jadechoghari/Ferret-UI-Llama8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jadechoghari/Ferret-UI-Llama8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jadechoghari/Ferret-UI-Llama8b",
		"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/jadechoghari/Ferret-UI-Llama8b

SGLang

How to use jadechoghari/Ferret-UI-Llama8b 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 "jadechoghari/Ferret-UI-Llama8b" \
    --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": "jadechoghari/Ferret-UI-Llama8b",
		"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 "jadechoghari/Ferret-UI-Llama8b" \
        --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": "jadechoghari/Ferret-UI-Llama8b",
		"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 jadechoghari/Ferret-UI-Llama8b with Docker Model Runner:
```
docker model run hf.co/jadechoghari/Ferret-UI-Llama8b
```

Clarification on Box Coordinates Scaling

by demitsuki - opened Nov 16, 2024

Discussion

demitsuki

Nov 16, 2024

Hi Ferret Team,

Thanks for sharing this checkpoint!

I ran the following example on your demo, using Ferret-UI-Llama8b model and default parameters:
{
"id": 0,
"image": "appstore_reminders.png",
"image_h": 2532,
"image_w": 1170,
"conversations": [
{
"from": "human",
"value": "\nWhere is the Games Tab located?"
}
]
}

The response returned: Games Tab [[0, 906, 256, 965]]. However, this box doesn’t seem to align with the "Games Tab" in the image, whether scaled or unscaled.

Could you clarify the scaling logic applied to the box and how I should interpret it?

Thanks!
Demi

jadechoghari

Owner Nov 20, 2024

Hi @demitsuki sorry for the late reply!
You can check the scaling logic here: https://github.com/apple/ml-ferret/blob/main/ferretui/ferretui/eval/model_UI.py

To speed up:

# ratio
ratio_w = VOCAB_IMAGE_W * 1.0 / image_wdith
ratio_h = VOCAB_IMAGE_H * 1.0 /image_height
def get_bbox_coor(box, ratio_w, ratio_h):
    return box[0] * ratio_w, box[1] * ratio_h, box[2] * ratio_w, box[3] * ratio_h

BoyuNLP

Jan 13, 2025

•

edited Jan 13, 2025

an example:

from PIL import Image, ImageDraw

image_path = "temp_image.png"

box_coordinates = [195, 906, 402, 970]

image = Image.open(image_path)

image_width, image_height = image.size

adjusted_box = [
    int(box_coordinates[0] / 1000 * image_width),  
    int(box_coordinates[1] / 1000 * image_height),
    int(box_coordinates[2] / 1000 * image_width),  
    int(box_coordinates[3] / 1000 * image_height), 
]


draw = ImageDraw.Draw(image)

draw.rectangle(adjusted_box, outline="red", width=3)

image.show()

miketes

Apr 17, 2025

@jadechoghari great project. But I am having the same problem. The model is giving

{
"id": 0,
"image": "appstore_reminders.png",
"image_h": 2532,
"image_w": 1170,
"conversations": [
{
"from": "human",
"value": "\nWhere is the Games Tab located?"
}
]
}

The response returned: Games Tab [[0, 906, 256, 965]]. However, this box doesn’t seem to align with the "Games Tab" in the image, whether scaled or unscaled.

How do you scale it. Because the coordinate is wrong.

Thank you

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