Instructions to use emirozcan16/AA-VL-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use emirozcan16/AA-VL-4B with Transformers:

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

pipe = pipeline("image-text-to-text", model="emirozcan16/AA-VL-4B")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("emirozcan16/AA-VL-4B")
model = AutoModelForMultimodalLM.from_pretrained("emirozcan16/AA-VL-4B", 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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use emirozcan16/AA-VL-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "emirozcan16/AA-VL-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "emirozcan16/AA-VL-4B",
		"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/emirozcan16/AA-VL-4B

SGLang

How to use emirozcan16/AA-VL-4B 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 "emirozcan16/AA-VL-4B" \
    --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": "emirozcan16/AA-VL-4B",
		"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 "emirozcan16/AA-VL-4B" \
        --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": "emirozcan16/AA-VL-4B",
		"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 emirozcan16/AA-VL-4B with Docker Model Runner:
```
docker model run hf.co/emirozcan16/AA-VL-4B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

AA-VL-4B — Data-Efficient GUI Grounding (4B)

Weights for AA-VL-4B: a LoRA (r=32) merged into Qwen3-VL-4B-Instruct for computer-screen grounding — a natural-language instruction maps to exact click coordinates on a screenshot.

Repo: emirozcan16/AA-VL-4B (merged safetensors)
GGUF for llama.cpp: emirozcan16/AA-VL-4B-GGUF
Base: Qwen/Qwen3-VL-4B-Instruct
Trained on ~7K curated samples: LoRA (r=32), initial adapter trained in 4-bit (QLoRA), final 250 steps in fp16. The assembled training set is not released for now (sources below); evaluation is fully open.
Interactive demo: AA-VL-4B-demo — real OSWorld-G predictions with the click and target region overlaid.

Scores (raw greedy, pass@1, single prediction, no crop/zoom/planner — our own runs; eval script + per-sample predictions included)

Benchmark	AA-VL-4B
OSWorld-G Original	65.2
OSWorld-G Refined	74.5
UI-Vision Final	29.93

UI-Vision breakdown: Basic 39.28 · Functional 37.75 · Spatial 14.21.

No cross-model comparison table here on purpose: published numbers for other models vary with prompt, protocol and agent scaffolding, so we report only our own raw runs and ship the artifacts to reproduce them.

Verification

Per-sample re-scoring artifacts ship with this repo — every raw generation, coordinate and ground-truth box is included:

Benchmark	File	Rows	Correct
OSWorld-G (Original / Refined)	`osworld_g_predictions.jsonl`	1128 (564 × 2)	368 / 420
UI-Vision Final	`predictions.jsonl`	5479	1640

python3 -c "import json;r=[json.loads(l) for l in open('osworld_g_predictions.jsonl')];print({v:sum(a['correct'] for a in r if a['variant']==v) for v in ('original','refined')})"

OSWorld-G rows store both the emitted 0–999 coordinates (pred_xy_norm999) and the pixel coordinates used by the official rule (pred_xy_px); UI-Vision rows store pred_xy_px + target_bbox_px. Every row carries the model output, the ground-truth box and the computed correct, so both benchmarks can be re-scored from scratch. See VERIFY.md.

Usage (transformers)

from transformers import AutoProcessor, AutoModelForImageTextToText
from PIL import Image

MODEL = "emirozcan16/AA-VL-4B"
proc = AutoProcessor.from_pretrained(MODEL, trust_remote_code=True)
model = AutoModelForImageTextToText.from_pretrained(
    MODEL, trust_remote_code=True, torch_dtype="auto", device_map="auto")

img = Image.open("screen.png").convert("RGB")
msgs = [{"role": "user", "content": [
    {"type": "image", "image": img},
    {"type": "text",
     "text": "Instruction: Click the funnel icon right of the search bar. "
             "Reply with the single action line now, nothing else:"}]}]
txt = proc.apply_chat_template(msgs, tokenize=False, add_generation_prompt=True)
inp = proc(text=[txt], images=[img], return_tensors="pt").to(model.device)
out = model.generate(**inp, max_new_tokens=64, do_sample=False)
print(proc.batch_decode(out[:, inp["input_ids"].shape[1]:],
                        skip_special_tokens=True)[0])
# -> click x=<int> y=<int>   (0-999 normalized scale)   or   refuse

Optional LoRA-only weights: emirozcan16/AA-VL-4B-adapter.

For the exact prompt used in the reported scores, see SYS_CLICK in osworld_g_eval.py; the snippet above is the minimal form.

Training data

~~7K curated rows (~~60% click, ~40% refusal). Sources:

Source	What we used	Role	License
Jedi (xlangai/Jedi, HF)	icon/component grounding clicks (train_raw/val_raw, os_layout, component libraries)	click labels	see upstream
GroundCUA	GUI grounding click slice (~2.5K rows, picked_json + images)	click labels	MIT
ScreenSpot-v2 (OS-Copilot/ScreenSpot-v2, HF)	desktop screenshots only	build synthetic refusal pairs	see upstream
Our synthesis	instruction ↔️ unrelated screenshot cross-pairing (doc–doc, icon–icon, desktop)	refusal examples	ours

Base model: Qwen/Qwen3-VL-4B-Instruct (Apache-2.0). No private or scraped user data. The assembled training set is not released for now. We did not train on any benchmark's answer labels.

Eval protocol

OSWorld-G: official point-in-box (510 click + 54 refusal), Original instructions.
Refined: official OSWorld-G_refined.json instructions, same 564 samples (510 click + 54 refusal), reported separately.
UI-Vision: official harness over its three subsets.

Single greedy prediction per sample, no crop/zoom/planner/pass@k.

Notes & limitations

Coordinates are on a 0–999 normalized grid.
Benchmarks were not independently reproduced by third parties; single seed.

Community quantizations

Thanks to mradermacher for independent static GGUF quants of this model — a big help for the llama.cpp community.

Project

Built in ~3 months on a ~$8 compute budget. Contact: ozcannemir08@gmail.com

Downloads last month: 40

Safetensors

Model size

4B params

Tensor type

F16

Model tree for emirozcan16/AA-VL-4B

Base model

Qwen/Qwen3-VL-4B-Instruct

Finetuned

(493)

this model

Quantizations

1 model

emirozcan16
/

AA-VL-4B