Instructions to use SparseLLM/DECO-0.2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SparseLLM/DECO-0.2B with Transformers:

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

pipe = pipeline("text-generation", model="SparseLLM/DECO-0.2B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("SparseLLM/DECO-0.2B", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use SparseLLM/DECO-0.2B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SparseLLM/DECO-0.2B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SparseLLM/DECO-0.2B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SparseLLM/DECO-0.2B

SGLang

How to use SparseLLM/DECO-0.2B 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 "SparseLLM/DECO-0.2B" \
    --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": "SparseLLM/DECO-0.2B",
		"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 "SparseLLM/DECO-0.2B" \
        --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": "SparseLLM/DECO-0.2B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SparseLLM/DECO-0.2B with Docker Model Runner:
```
docker model run hf.co/SparseLLM/DECO-0.2B
```

DECO-0.2B / README.md

Raincleared

Update README.md

b9a3cf5 verified about 18 hours ago

preview code

raw

history blame contribute delete

2.14 kB

	---
	language:
	- en
	- zh
	license: apache-2.0
	pipeline_tag: text-generation
	library_name: transformers
	---

	# DECO-0.2B

	This is the 0.2B DECO checkpoint introduced by the paper [DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices](https://huggingface.co/papers/2605.10933).

	DECO (Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices) is a sparse MoE architecture designed to match the performance of dense Transformers under identical total parameter budgets and training tokens. It is an improved version of the [BlockFFN](https://arxiv.org/pdf/2507.08771) architecture.

	- Authors: Chenyang Song, Weilin Zhao, Xu Han, Chaojun Xiao, Yingfa Chen, Zhiyuan Liu
	- Paper: [arXiv:2605.10933](https://huggingface.co/papers/2605.10933)
	- Code: [https://github.com/thunlp/DECO](https://github.com/thunlp/DECO)

	### Quick start

	You can load and use this model with `AutoTokenizer` and `AutoModelForCausalLM` from `transformers`. Since the model uses a custom architecture, `trust_remote_code=True` is required.

	```python
	import torch
	from transformers import AutoTokenizer, AutoModelForCausalLM

	model_id = "SparseLLM/DECO-0.2B"
	tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
	model = AutoModelForCausalLM.from_pretrained(
	model_id,
	torch_dtype=torch.bfloat16,
	trust_remote_code=True,
	).to("cuda").eval()

	prompt = "Mixture-of-Experts models are useful because"
	inputs = tokenizer(prompt, return_tensors="pt").to("cuda")

	with torch.no_grad():
	output = model.generate(**inputs, max_new_tokens=64, do_sample=False)

	print(tokenizer.decode(output[0], skip_special_tokens=True))
	```

	### Citation
	If you find our work useful for your research, please kindly cite our paper as follows:

	```bibtex
	@article{song2026deco,
	title={{DECO}: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices},
	author={Chenyang Song, Weilin Zhao, Xu Han, Chaojun Xiao, Yingfa Chen, Zhiyuan Liu},
	journal={arXiv preprint arXiv:2605.10933},
	year={2026},
	url={https://arxiv.org/pdf/2605.10933},
	}
	```