Instructions to use inclusionAI/Ring-mini-sparse-2.0-exp with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use inclusionAI/Ring-mini-sparse-2.0-exp with Transformers:

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

pipe = pipeline("text-generation", model="inclusionAI/Ring-mini-sparse-2.0-exp", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("inclusionAI/Ring-mini-sparse-2.0-exp", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use inclusionAI/Ring-mini-sparse-2.0-exp with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "inclusionAI/Ring-mini-sparse-2.0-exp"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "inclusionAI/Ring-mini-sparse-2.0-exp",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/inclusionAI/Ring-mini-sparse-2.0-exp

SGLang

How to use inclusionAI/Ring-mini-sparse-2.0-exp 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 "inclusionAI/Ring-mini-sparse-2.0-exp" \
    --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": "inclusionAI/Ring-mini-sparse-2.0-exp",
		"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 "inclusionAI/Ring-mini-sparse-2.0-exp" \
        --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": "inclusionAI/Ring-mini-sparse-2.0-exp",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use inclusionAI/Ring-mini-sparse-2.0-exp with Docker Model Runner:
```
docker model run hf.co/inclusionAI/Ring-mini-sparse-2.0-exp
```

Config issue

by wsbagnsv1 - opened Oct 21, 2025

Discussion

wsbagnsv1

Oct 21, 2025

When i tried to run the new Ring-mini-sparse-2.0-exp i got this error "AttributeError: 'BailingMoeV2Config' object has no attribute 'moe_router_topk_scaling_factor'"
If i just set that to 1.0 i get "AttributeError: 'BailingMoeV2Config' object has no attribute 'use_expert_bias'" next so the config seems to be broken?
Also even when i set dummy values i get the message that some tensors are missing in the model?
So if im not mistaken that seems to be a config.json issue right?

bestfleer

inclusionAI org Oct 22, 2025

We have fixed both configuration_bailing_moe_v2.py and modeling_bailing_moe_v2.py. The model can now be properly loaded and used for inference via Hugging Face.
However, please note that sparse attention is not supported during the decoding phase in the current HF Transformers implementation.

To fully leverage the sparse attention mechanism—especially during decoding—we recommend using SGLang, as detailed in our README.

wsbagnsv1

Oct 22, 2025

thank you very much (;

wsbagnsv1 changed discussion status to closed Oct 22, 2025

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment