Instructions to use harrrshall/lumen-merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use harrrshall/lumen-merged with Transformers:

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

pipe = pipeline("text-generation", model="harrrshall/lumen-merged")
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("harrrshall/lumen-merged")
model = AutoModelForMultimodalLM.from_pretrained("harrrshall/lumen-merged", 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 harrrshall/lumen-merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "harrrshall/lumen-merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "harrrshall/lumen-merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/harrrshall/lumen-merged

SGLang

How to use harrrshall/lumen-merged 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 "harrrshall/lumen-merged" \
    --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": "harrrshall/lumen-merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "harrrshall/lumen-merged" \
        --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": "harrrshall/lumen-merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use harrrshall/lumen-merged with Docker Model Runner:
```
docker model run hf.co/harrrshall/lumen-merged
```

Lumen estimated ~200× lower inference cost with frontier competitive saferepair

Lumen is a 4b model for automatically verifiable accessibility attribute repair in react and static html. on a matched 43-task development benchmark, Lumen completed 42 safe repairs, compared with 40 for gpt-5.6 sol and 43 for claude opus 4.8.

this is the standalone bf16 edition of Lumen. it includes the qwen3.5 4b base weights with the Lumen lora adapter merged in, so most users need to download only this repository.

the smaller harrrshall/lumen repository is the 252 mb lora adapter. use that edition when you already store the exact qwen base or want to manage adapters yourself.

quick start

import torch
from transformers import AutoModelForImageTextToText, AutoProcessor

model_id = "harrrshall/lumen-merged"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    dtype=torch.bfloat16,
    device_map="auto",
)
model.eval()

no peft adapter attachment and no separate base model download are required. the repository is approximately 9.1 gb before local filesystem overhead.

what Lumen does

Lumen generates small structured source patches for automatically verifiable accessibility defects in react and static html. its intended workflow rebuilds the application and validates the patch with browser, keyboard, semantic, visual, regression, and protected path gates.

frontier comparison

system	tasks	saferepair	cost per safe repair
Lumen	43 tasks	42/43 (97.67%)	$0.000268
gpt-5.6 sol	43 tasks	40/43 (93.02%)	$0.070857
claude opus 4.8	43 tasks	43/43 (100%)	$0.054012

provenance

base model: Qwen/Qwen3.5-4B
pinned base revision: 851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a
post training: qlora supervised fine tuning
merged checkpoint dtype: bf16
original adapter: harrrshall/lumen

training curve

limitations

Lumen covers deterministic source repair families supported by its evaluator. it does not establish complete wcag or legal conformance. review every patch before merging, and involve qualified accessibility practitioners and assistive technology users for subjective decisions.

Downloads last month: -

Safetensors

Model size

5B params

Tensor type

BF16

Model tree for harrrshall/lumen-merged

Base model

Qwen/Qwen3.5-4B-Base

Finetuned

Qwen/Qwen3.5-4B

Finetuned

(449)

this model