Instructions to use quelmap/Lightning-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use quelmap/Lightning-4b with Transformers:

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

pipe = pipeline("text-generation", model="quelmap/Lightning-4b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("quelmap/Lightning-4b", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use quelmap/Lightning-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "quelmap/Lightning-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": "quelmap/Lightning-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/quelmap/Lightning-4b

SGLang

How to use quelmap/Lightning-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 "quelmap/Lightning-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": "quelmap/Lightning-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 "quelmap/Lightning-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": "quelmap/Lightning-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use quelmap/Lightning-4b with Docker Model Runner:
```
docker model run hf.co/quelmap/Lightning-4b
```

Lightning-4b / README.md

shintaro-29

Update README.md

36b2b39 verified 11 months ago

preview code

Raw

History Blame Contribute Delete

2.59 kB

	---
	license: apache-2.0
	language:
	- en
	base_model:
	- Qwen/Qwen3-4B-Thinking-2507
	pipeline_tag: text-generation
	library_name: transformers
	---
	# Lightning-4b - Your Local data analysis agent

	[![GitHub](https://img.shields.io/badge/GitHub-Repository-blue?logo=github)](https://github.com/quelmap-inc/quelmap)
	[![License](https://img.shields.io/badge/License-Apache%202.0-yellow)](https://opensource.org/licenses/Apache-2.0)

	![image/svg](.assets/lightning-logo2.svg)

	## Overview
	Lightning-4b is a language model specifically designed and trained for data analysis tasks on local devices. With just a laptop (fully tested on an M4 MacBook Air with 16GB RAM), you can process data without ever sending it to major LLM provider.
	### What it can do
	- Data visualization
	- Table joins
	- t-tests
	- Unlimited rows, 30+ tables analyzed simultaneously
	### What it cannot do
	- Business reasoning or management decision-making advice
	- Multi-turn analysis

	To use this model, install [quelmap](https://github.com/quelmap-inc/quelmap) on your device.
	Quelmap is an open-source data analysis assistant with every essential features like data upload and an built-in python sandbox.
	For installation instructions, see the [Quick Start](https://quelmap.com/quickstart).
	![image/gif](.assets/preview.gif)
	### Performance
	This model was trained specifically for use with [quelmap](https://github.com/quelmap-inc/quelmap).
	It was evaluated using a sample database and 122 analysis queries, and achieved performance surpassing models with 50x more parameters.

	For details about the model and its training process, see the [Lightning-4b Details](https://quelmap.com/lightning-4b) page.

	![image/svg](.assets/lightning.svg)

	### Running Model on your machine
	You can easily install Lightning-4b and quelmap by following the [Quick Start](https://quelmap.com/quickstart).

	Lightning-4b has multiple quantization versions depending on your hardware.
	It runs smoothly on laptops, and on higher-spec machines it can handle more tables (30+ tables) and longer chat histories.

	Example Specs and Model Versions
	- Laptop (e.g. mac book air 16GB) - 4bit Quantization + 10,240 Context Window
	```
	ollama pull hf.co/quelmap/Lightning-4b-GGUF-short-ctx:Q4_K_M
	```
	- Gaming Laptop - 4bit Quantization + 40,960 Context Window
	```
	ollama pull hf.co/quelmap/Lightning-4b-GGUF:Q4_K_M
	```
	- Powerful PC with GPU - No Quantization + 40,960 Context Window
	```
	ollama pull hf.co/quelmap/Lightning-4b-GGUF:F16
	```

	For more details, please refer to the [Lightning-4b Details](https://quelmap.com/lightning-4b) page.