Instructions to use chiennv/Orthrus-Qwen3-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use chiennv/Orthrus-Qwen3-4B with Transformers:

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

pipe = pipeline("text-generation", model="chiennv/Orthrus-Qwen3-4B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("chiennv/Orthrus-Qwen3-4B", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("chiennv/Orthrus-Qwen3-4B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use chiennv/Orthrus-Qwen3-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "chiennv/Orthrus-Qwen3-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "chiennv/Orthrus-Qwen3-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/chiennv/Orthrus-Qwen3-4B

SGLang

How to use chiennv/Orthrus-Qwen3-4B 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 "chiennv/Orthrus-Qwen3-4B" \
    --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": "chiennv/Orthrus-Qwen3-4B",
		"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 "chiennv/Orthrus-Qwen3-4B" \
        --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": "chiennv/Orthrus-Qwen3-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use chiennv/Orthrus-Qwen3-4B with Docker Model Runner:
```
docker model run hf.co/chiennv/Orthrus-Qwen3-4B
```

chiennv commited on 1 day ago

Commit

fdbc127

verified ·

1 Parent(s): 694a49d

Update README.md

Browse files

Files changed (1) hide show

README.md +15 -2

README.md CHANGED Viewed

@@ -11,7 +11,7 @@ pipeline_tag: text-generation
 ---
 # Orthrus-Qwen3-4B
-[**Paper**](https://arxiv.org) | [**GitHub**](https://github.com/chiennv2000/orthrus)
 **Orthrus** is a simple and efficient dual-architecture framework that unifies the exact generation fidelity of autoregressive Large Language Models (LLMs) with the high-speed parallel token generation of diffusion models. By augmenting a frozen pre-trained LLM with a lightweight, trainable diffusion module, Orthrus delivers significantly accelerated inference without sacrificing output quality.
@@ -77,4 +77,17 @@ output_ids = model.generate(
 ```
 ## Citation
-If you find this model or architecture useful in your work, please cite the original paper:

 ---
 # Orthrus-Qwen3-4B
+[**Paper**](https://arxiv.org/abs/2605.12825) | [**GitHub**](https://github.com/chiennv2000/orthrus)
 **Orthrus** is a simple and efficient dual-architecture framework that unifies the exact generation fidelity of autoregressive Large Language Models (LLMs) with the high-speed parallel token generation of diffusion models. By augmenting a frozen pre-trained LLM with a lightweight, trainable diffusion module, Orthrus delivers significantly accelerated inference without sacrificing output quality.
 ```
 ## Citation
+If you find this model or architecture useful in your work, please cite the original paper:
+```bibtex
+@misc{vannguyen2026orthrusmemoryefficientparalleltoken,
+      title={Orthrus: Memory-Efficient Parallel Token Generation via Dual-View Diffusion},
+      author={Chien Van Nguyen and Chaitra Hegde and Van Cuong Pham and Ryan A. Rossi and Franck Dernoncourt and Thien Huu Nguyen},
+      year={2026},
+      eprint={2605.12825},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG},
+      url={https://arxiv.org/abs/2605.12825},
+}
+```