Instructions to use drunksu/GUISwiper with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use drunksu/GUISwiper with Transformers:

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

pipe = pipeline("image-text-to-text", model="drunksu/GUISwiper")
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("drunksu/GUISwiper")
model = AutoModelForMultimodalLM.from_pretrained("drunksu/GUISwiper", 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 drunksu/GUISwiper with vLLM:

Install from pip and serve model

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

SGLang

How to use drunksu/GUISwiper 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 "drunksu/GUISwiper" \
    --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": "drunksu/GUISwiper",
		"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 "drunksu/GUISwiper" \
        --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": "drunksu/GUISwiper",
		"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 drunksu/GUISwiper with Docker Model Runner:
```
docker model run hf.co/drunksu/GUISwiper
```

drunksu commited on 9 days ago

Commit

41c488c

verified ·

1 Parent(s): 461b647

Update README.md

Browse files

Files changed (1) hide show

README.md +5 -9

README.md CHANGED Viewed

@@ -1,8 +1,6 @@
 ---
 license: apache-2.0
 base_model: Qwen/Qwen2.5-VL-3B-Instruct
-datasets:
-- TSKGHS17/SwipeBench
 language:
 - en
 - zh
@@ -16,15 +14,14 @@ tags:
 - reinforcement-learning
 ---
-# GUISwiper 3B (RL)
 GUISwiper is an RL-aligned GUI agent model for **human-like swipe synthesis**, introduced in the paper
 [SwipeGen: Bridging the Execution Gap in GUI Agents via Human-like Swipe Synthesis](https://arxiv.org/abs/2601.18305)
 (ACM MM 2026 Oral).
 This repository hosts the **final RL-aligned 3B checkpoint** (bfloat16), fine-tuned from
-[`Qwen/Qwen2.5-VL-3B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct) on
-[SwipeBench](https://huggingface.co/datasets/TSKGHS17/SwipeBench).
 ## Model Details
@@ -33,7 +30,7 @@ This repository hosts the **final RL-aligned 3B checkpoint** (bfloat16), fine-tu
 | Base model      | Qwen/Qwen2.5-VL-3B-Instruct                  |
 | Parameters      | 3B (bfloat16, ~7 GB)                         |
 | Training        | SFT + RL alignment on SwipeBench             |
-| Input           | GUI screenshots / screen videos + instruction |
 | Output          | Human-like swipe action (trajectory)        |
 | Hardware        | NVIDIA GPUs (see paper for details)          |
@@ -57,7 +54,7 @@ processor = AutoProcessor.from_pretrained(repo_id)
 image = load_your_gui_screenshot()  # PIL.Image
 messages = [{"role": "user", "content": [
     {"type": "image", "image": image},
-    {"type": "text", "text": "Describe the swipe to perform here."},
 ]}]
 text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
 inputs = processor(text=[text], images=[image], return_tensors="pt").to(model.device)
@@ -91,10 +88,9 @@ path = hf_hub_download("drunksu/GUISwiper", "model-00001-of-00002.safetensors")
 - Paper: https://arxiv.org/abs/2601.18305
 - Project / code: https://github.com/TSKGHS17/SwipeGen
-- Dataset (SwipeBench): https://huggingface.co/datasets/TSKGHS17/SwipeBench
 ## License & Disclaimer
 The model weights are released under Apache-2.0, consistent with the base model
 `Qwen2.5-VL-3B-Instruct`. Users should comply with the original license terms of Qwen2.5-VL
-and use the model responsibly; outputs are generated by AI and may contain errors.

 ---
 license: apache-2.0
 base_model: Qwen/Qwen2.5-VL-3B-Instruct
 language:
 - en
 - zh
 - reinforcement-learning
 ---
+# GUISwiper
 GUISwiper is an RL-aligned GUI agent model for **human-like swipe synthesis**, introduced in the paper
 [SwipeGen: Bridging the Execution Gap in GUI Agents via Human-like Swipe Synthesis](https://arxiv.org/abs/2601.18305)
 (ACM MM 2026 Oral).
 This repository hosts the **final RL-aligned 3B checkpoint** (bfloat16), fine-tuned from
+[`Qwen/Qwen2.5-VL-3B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct).
 ## Model Details
 | Base model      | Qwen/Qwen2.5-VL-3B-Instruct                  |
 | Parameters      | 3B (bfloat16, ~7 GB)                         |
 | Training        | SFT + RL alignment on SwipeBench             |
+| Input           | GUI screenshots + instruction |
 | Output          | Human-like swipe action (trajectory)        |
 | Hardware        | NVIDIA GPUs (see paper for details)          |
 image = load_your_gui_screenshot()  # PIL.Image
 messages = [{"role": "user", "content": [
     {"type": "image", "image": image},
+    {"type": "text", "text": "Swipe to learn more."},
 ]}]
 text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
 inputs = processor(text=[text], images=[image], return_tensors="pt").to(model.device)
 - Paper: https://arxiv.org/abs/2601.18305
 - Project / code: https://github.com/TSKGHS17/SwipeGen
 ## License & Disclaimer
 The model weights are released under Apache-2.0, consistent with the base model
 `Qwen2.5-VL-3B-Instruct`. Users should comply with the original license terms of Qwen2.5-VL
+and use the model responsibly; outputs are generated by AI and may contain errors.