Instructions to use ministral/Ministral-3b-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ministral/Ministral-3b-instruct with Transformers:

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

pipe = pipeline("text-generation", model="ministral/Ministral-3b-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ministral/Ministral-3b-instruct")
model = AutoModelForCausalLM.from_pretrained("ministral/Ministral-3b-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ministral/Ministral-3b-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ministral/Ministral-3b-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ministral/Ministral-3b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ministral/Ministral-3b-instruct

SGLang

How to use ministral/Ministral-3b-instruct 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 "ministral/Ministral-3b-instruct" \
    --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": "ministral/Ministral-3b-instruct",
		"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 "ministral/Ministral-3b-instruct" \
        --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": "ministral/Ministral-3b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ministral/Ministral-3b-instruct with Docker Model Runner:
```
docker model run hf.co/ministral/Ministral-3b-instruct
```

Training, fine-tuning instructions

by Muhammadreza - opened Mar 8, 2025

Discussion

Muhammadreza

Mar 8, 2025

Greetings. Is there any documents or guides on fine tuning this model?

gmonsoon

Ministral org Mar 11, 2025

Hello, You need to use mergekit to distill the pretrained model, for example mistral or other model, by remove the layers of the model.
then, just restart pretraining and instruct tuning the distilled model.

You can either make smaller or bigger parameters by using Mergekit.

Muhammadreza

Mar 13, 2025

Thank you @gmonsoon . I will give it a shot.

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