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

Libraries

How to use inclusionAI/Ling-lite with Transformers:

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

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

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use inclusionAI/Ling-lite with vLLM:

Install from pip and serve model

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

Use Docker

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

SGLang

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

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

zhujiangang commited on Apr 18, 2025

Commit

ed7d815

verified ·

1 Parent(s): 8cd99d3

Update README.md

Browse files

Files changed (1) hide show

README.md +23 -0

README.md CHANGED Viewed

@@ -24,6 +24,27 @@ Their structure makes it easy to scale up and down and adapt to different tasks,
 As more developers and researchers engage with the platform, we can expect rapid advancements and improvements, leading to even more sophisticated applications. This collaborative approach accelerates development and ensures that the models remain at the forefront of technology, addressing emerging challenges in various fields.
 ## Model Downloads
 You can download the following table to see the various parameters for your use case. If you are located in mainland China, we also provide the model on ModelScope.cn to speed up the download process.
@@ -36,6 +57,8 @@ You can download the following table to see the various parameters for your use
 | Ling-lite |       16.8B       |         2.75B         |        64K         |     [🤗 HuggingFace](https://huggingface.co/inclusionAI/Ling-lite)|
 </div>
 ## Evaluation
 Detailed evaluation results are reported in our [technical report](https://github.com/inclusionAI/Ling/blob/0eb5350635c5a80b7663eeb3412f181fe0c91675/Ling_TR_v1.pdf).

 As more developers and researchers engage with the platform, we can expect rapid advancements and improvements, leading to even more sophisticated applications. This collaborative approach accelerates development and ensures that the models remain at the forefront of technology, addressing emerging challenges in various fields.
+## Update
+Ling-lite is upgraded to Ling-lite-0415. The new model demonstrates notable improvements over its predecessor, Ling-lite-0220, especially on code and math.
+<div align="center">
+|      **Benchmark**       |   **#shots**       | **Ling-Lite-0415** | **Ling-Lite-0220** | **Qwen2.5-7B-Instruct** |  **LLaMA3.1-8B** |
+| :------------------: | :------------------: | :---------------: | :-------------------: | :----------------: | :----------------: |
+|    MMLU(EM)    | 5 |   74.87      |      71.27            |       74.26         |  68.67  |
+|      GPQA(Pass@1)       | 0 |  40.91     |         28.66         |       34.47         |  32.80  |
+|    HumanEval(Pass@1)    | 0 |  89.02     |        83.54          |       87.20         |  70.73  |
+|      LiveCodeBench 2408-2411 (Pass@1)      | 0 |    24.11     |        15.18          |       16.96         |  11.61  |
+| LCBench(pass@1) | 0 |   60.00      |        47.22          |      54.17          |  29.04  |
+|   Math(EM)    | 0 |    79.12     |        72.80          |      73.66          |  52.42  |
+|   AIME2024(pass@1)    | 0 |    13.33     |        6.67          |      16.67          |  0.00  |
+|   OlympiadBench(pass@1)    | 0 |     37.33    |       34.42           |      37.19          |  16.3  |
+|   BBH(EM)    | 0 |    74.58     |        66.38          |         66.07       |  68.05  |
+|   IFEval(Prompt Strict)   | 0 |    81.09     |        77.99          |       71.16         |  53.45  |
+</div>
 ## Model Downloads
 You can download the following table to see the various parameters for your use case. If you are located in mainland China, we also provide the model on ModelScope.cn to speed up the download process.
 | Ling-lite |       16.8B       |         2.75B         |        64K         |     [🤗 HuggingFace](https://huggingface.co/inclusionAI/Ling-lite)|
 </div>
+Note: Ling-lite has been upgrade to Ling-lite-0415. The previous version, Ling-lite-0220, can be found in branch `ling-lite-0220`.
 ## Evaluation
 Detailed evaluation results are reported in our [technical report](https://github.com/inclusionAI/Ling/blob/0eb5350635c5a80b7663eeb3412f181fe0c91675/Ling_TR_v1.pdf).