Instructions to use mobileforge-anonymous/ForgeOwl-8B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mobileforge-anonymous/ForgeOwl-8B with Transformers:

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

pipe = pipeline("image-text-to-text", model="mobileforge-anonymous/ForgeOwl-8B")
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 AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("mobileforge-anonymous/ForgeOwl-8B")
model = AutoModelForMultimodalLM.from_pretrained("mobileforge-anonymous/ForgeOwl-8B", device_map="auto")
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?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use mobileforge-anonymous/ForgeOwl-8B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mobileforge-anonymous/ForgeOwl-8B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mobileforge-anonymous/ForgeOwl-8B",
		"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/mobileforge-anonymous/ForgeOwl-8B

SGLang

How to use mobileforge-anonymous/ForgeOwl-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 "mobileforge-anonymous/ForgeOwl-8B" \
    --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": "mobileforge-anonymous/ForgeOwl-8B",
		"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 "mobileforge-anonymous/ForgeOwl-8B" \
        --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": "mobileforge-anonymous/ForgeOwl-8B",
		"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 mobileforge-anonymous/ForgeOwl-8B with Docker Model Runner:
```
docker model run hf.co/mobileforge-anonymous/ForgeOwl-8B
```

ForgeOwl-8B / README.md

mobileforge-anonymous

Publish sanitized anonymous files

20616ae verified 1 day ago

preview code

Raw

History Blame Contribute Delete

1.73 kB

	---
	license: mit
	base_model: mPLUG/GUI-Owl-1.5-8B-Instruct
	library_name: transformers
	pipeline_tag: image-text-to-text
	tags:
	- mobile-gui
	- vision-language-model
	- anonymous-submission
	---

	# ForgeOwl-8B

	ForgeOwl-8B is the GUI-Owl-1.5-8B-Instruct policy adapted with MobileForge on `900` automatically generated target-app tasks. MobileForge uses the policy's own rollouts, hierarchical critic feedback, corrective hints, and hint-contextualized step-level GRPO. No human-written adaptation tasks, demonstrations, or reward labels are used.

	This repository is part of an anonymous ICLR submission artifact. Author and paper-identifying metadata will be added after review.

	Anonymous project page: https://mobileforge-anonymous.github.io/

	## Evaluation

	On AndroidWorld (`116` tasks), this checkpoint obtains `78/116` (`67.2%`) Pass@1, `87/116` (`75.0%`) Pass@2, and `90/116` (`77.6%`) Pass@3. On the MobileWorld GUI-only split (`117` tasks), it obtains `48/117` (`41.0%`) success rate.

	Reproduction code and raw evaluation archives are available from the [anonymous code repository](https://github.com/mobileforge-anonymous/MobileForge) and [benchmark-results dataset](https://huggingface.co/datasets/mobileforge-anonymous/mobileforge-benchmark-results).

	## Usage

	Use the same loading and prompting interface as [`mPLUG/GUI-Owl-1.5-8B-Instruct`](https://huggingface.co/mPLUG/GUI-Owl-1.5-8B-Instruct). See the anonymous code repository for AndroidWorld and MobileWorld runners.

	## Limitations

	The model can make incorrect or unsafe GUI actions. Run it only in isolated test environments, inspect actions before using it with personal data, and do not treat benchmark success as evidence of general reliability.