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

Libraries

How to use inclusionAI/Ring-lite with Transformers:

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

pipe = pipeline("text-generation", model="inclusionAI/Ring-lite", 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-lite", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use inclusionAI/Ring-lite with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "inclusionAI/Ring-lite"
# 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-lite",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/inclusionAI/Ring-lite

SGLang

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

Docker Model Runner
How to use inclusionAI/Ring-lite with Docker Model Runner:
```
docker model run hf.co/inclusionAI/Ring-lite
```

LiangJiang commited on Jul 4, 2025

Commit

5edf18f

verified ·

1 Parent(s): 091736d

Update README.md

Browse files

Files changed (1) hide show

README.md +1 -0

README.md CHANGED Viewed

@@ -64,6 +64,7 @@ More details are reported in our [technical report](https://arxiv.org/abs/2506.1
 The newly updated **Ring-lite** model now supports two distinct reasoning modes: "**thinking on**" and "**thinking off**". These modes are controlled by the `enable_thinking` parameter in the `tokenizer.apply_chat_template()` function.
 * When `enable_thinking` is set to `True` (or omitted), the model operates in "**thinking on**" mode, where it generates and outputs the internal reasoning process.
 * When `enable_thinking` is explicitly set to `False`, the model runs in "**thinking off**" mode, skipping the reasoning step entirely and directly producing the final answer.
 This feature allows users to choose between detailed reasoning and concise output based on their specific needs.
 #### Thinking on

 The newly updated **Ring-lite** model now supports two distinct reasoning modes: "**thinking on**" and "**thinking off**". These modes are controlled by the `enable_thinking` parameter in the `tokenizer.apply_chat_template()` function.
 * When `enable_thinking` is set to `True` (or omitted), the model operates in "**thinking on**" mode, where it generates and outputs the internal reasoning process.
 * When `enable_thinking` is explicitly set to `False`, the model runs in "**thinking off**" mode, skipping the reasoning step entirely and directly producing the final answer.
 This feature allows users to choose between detailed reasoning and concise output based on their specific needs.
 #### Thinking on