Instructions to use OraRL/Video-ORA-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use OraRL/Video-ORA-4B with Transformers:

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

pipe = pipeline("image-text-to-text", model="OraRL/Video-ORA-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)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("OraRL/Video-ORA-4B")
model = AutoModelForMultimodalLM.from_pretrained("OraRL/Video-ORA-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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use OraRL/Video-ORA-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "OraRL/Video-ORA-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": "OraRL/Video-ORA-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/OraRL/Video-ORA-4B

SGLang

How to use OraRL/Video-ORA-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 "OraRL/Video-ORA-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": "OraRL/Video-ORA-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 "OraRL/Video-ORA-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": "OraRL/Video-ORA-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 OraRL/Video-ORA-4B with Docker Model Runner:
```
docker model run hf.co/OraRL/Video-ORA-4B
```

Video-ORA-4B / code /docs /environment.md

lyhisme

Add 4B model card, benchmark assets, and release code

0185029 verified about 12 hours ago

preview code

Raw

History Blame Contribute Delete

1.82 kB

简体中文

Reproduce the OraRL Environment

OraRL uses one reproducible software stack for policy optimization and task-native evaluation. The released configuration was validated on NVIDIA H20/Hopper GPUs with CUDA 12.9 and includes both runtimes in this repository.

Requirements

Linux with Conda available in PATH
NVIDIA GPU with a recent driver
CUDA 12.9-compatible hardware for the paper-matched environment

CUDA 12.9 GA officially requires NVIDIA Linux driver 575.51.03 or newer. Other NVIDIA GPUs may work, but the published setup was validated on H20.

Install the pinned stack

bash scripts/create_conda_env.sh
conda activate orarl

python scripts/check_environment.py \
  --require-gpu \
  --model /path/to/local/model

The installer pins Python 3.11, PyTorch 2.10.0+cu129, Transformers 5.5.4, vLLM 0.19.1, FlashAttention 2.8.3, and the remaining packages in requirements-cu129.txt. The final editable install exposes the orarl-* commands and the bundled verl trainer package from this checkout; do not install an unrelated verl release from PyPI over it.

Validate an evaluation-only node

If a node will only run evaluation, skip the training-side checks:

bash scripts/create_conda_env.sh --evaluation-only
conda activate orarl

python scripts/check_environment.py --evaluation-only

Verify the release checkout

Run the release checks before allocating a long job:

python scripts/check_release.py
python -m pytest -q
ruff check .

On a paper-matched H20 node, add --require-gpu --require-h20 to scripts/check_environment.py. Use the same source revision and environment on every node of a distributed run.

Continue with Training or Evaluation.