OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis
[🏠Homepage] [💻Code] [📝Paper] [🤗Models][🤗Data]
Overview
We introduce OS-Genesis, an interaction-driven pipeline that synthesizes high-quality and diverse GUI agent trajectory data without human supervision. By leveraging reverse task synthesis, OS-Genesis enables effective training of GUI agents to achieve superior performance on dynamic benchmarks such as AndroidWorld and WebArena.
Quick Start
OS-Genesis-7B-AC is a mobile action model finetuned from Qwen2-VL-7B-Instruct.
OS-Genesis AC Family Models
In the following table, we provide an overview of the OS-Genesis AC Family Models used for evaluating the AndroidControl Benchmark.
| Model Name | Base Model | Training Data | HF Link |
|---|---|---|---|
| OS-Genesis-4B-AC | InternVL2-4B | OS-Genesis-ac-training-data | 🤗 link |
| OS-Genesis-7B-AC | Qwen2-VL-7B-Instruct | OS-Genesis-ac-training-data | 🤗 link |
| OS-Genesis-8B-AC | InternVL2-8B | OS-Genesis-ac-training-data | 🤗 link |
Inference Example
First, ensure that the necessary dependencies are installed:
pip install transformers
pip install qwen-vl-utils
For evaluating the AndroidControl Benchmark, please refer to the evaluation code.
Inference code example:
from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
from qwen_vl_utils import process_vision_info
# Default: Load the model on the available device(s)
model = Qwen2VLForConditionalGeneration.from_pretrained(
"OS-Copilot/OS-Genesis-7B-AC", torch_dtype="auto", device_map="auto"
)
processor = AutoProcessor.from_pretrained("OS-Copilot/OS-Atlas-Base-7B")
messages = [
{
"role": "user",
"content": [
{
"type": "image",
"image": "./web_6f93090a-81f6-489e-bb35-1a2838b18c01.png",
},
{"type": "text", "text": "You are a GUI task expert, I will provide you with a high-level instruction, an action history, a screenshot with its corresponding accessibility tree.\n High-level instruction: {high_level_instruction}\n Action history: {action_history}\n Accessibility tree: {a11y_tree}\n Please generate the low-level thought and action for the next step."},
],
}
]
# Preparation for inference
text = processor.apply_chat_template(
messages, tokenize=False, add_generation_prompt=True
)
image_inputs, video_inputs = process_vision_info(messages)
inputs = processor(
text=[text],
images=image_inputs,
videos=video_inputs,
padding=True,
return_tensors="pt",
)
inputs = inputs.to("cuda")
# Inference: Generation of the output
generated_ids = model.generate(**inputs, max_new_tokens=128)
generated_ids_trimmed = [
out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
generated_ids_trimmed, skip_special_tokens=False, clean_up_tokenization_spaces=False
)
print(output_text)
# <|object_ref_start|>language switch<|object_ref_end|><|box_start|>(576,12),(592,42)<|box_end|><|im_end|>
Citation
If you find this repository helpful, feel free to cite our paper:
@article{sun2024genesis,
title={OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesis},
author={Sun, Qiushi and Cheng, Kanzhi and Ding, Zichen and Jin, Chuanyang and Wang, Yian and Xu, Fangzhi and Wu, Zhenyu and Jia, Chengyou and Chen, Liheng and Liu, Zhoumianze and others},
journal={arXiv preprint arXiv:2412.19723},
year={2024}
}