Instructions to use Finisha-F-scratch/Mini-mistral-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Finisha-F-scratch/Mini-mistral-v1 with Transformers:

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

pipe = pipeline("text-generation", model="Finisha-F-scratch/Mini-mistral-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Finisha-F-scratch/Mini-mistral-v1")
model = AutoModelForCausalLM.from_pretrained("Finisha-F-scratch/Mini-mistral-v1")
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 Finisha-F-scratch/Mini-mistral-v1 with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Finisha-F-scratch/Mini-mistral-v1

SGLang

How to use Finisha-F-scratch/Mini-mistral-v1 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 "Finisha-F-scratch/Mini-mistral-v1" \
    --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": "Finisha-F-scratch/Mini-mistral-v1",
		"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 "Finisha-F-scratch/Mini-mistral-v1" \
        --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": "Finisha-F-scratch/Mini-mistral-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Finisha-F-scratch/Mini-mistral-v1 with Docker Model Runner:
```
docker model run hf.co/Finisha-F-scratch/Mini-mistral-v1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

💎 Mini-mistral-v1 : L'Essence du SLM 💎

Mini-mistral-v1 est une prouesse de miniaturisation de la forge Finisha. Avec seulement 49 millions de paramètres.

🛠️ Spécifications de la Forge

Base de Forge : Version affinée (fine-tuned) de ilyana-pretrain
Architecture : Causal LM optimisé pour la génération de texte fluide
Poids : 49M paramètres (Ultra-lightweight)
Philosophie : Un modèle "From Scratch" qui privilégie la structure de l'original au détriment du lissage industriel

✨ Caractéristiques Techniques

Efficacité Causal : Hérite de la capacité d'Ilyana à dérouler une syntaxe naturelle sans les lourdeurs des modèles massifs.
Zéro Alignement : Pas de filtres de politesse artificielle ; le modèle répond avec la texture brute de ses données d'entraînement

🌶️ attention a la confusion 🌶️

Mini-mistral n'a rien a voir avec le grand mistral de MistralAI, il est un SLM from scratch 100 % indépendant avec sa propre voix et entraîné pour avoir sa propre synthaxe et sa propre vision du monde dans la structure de ce qu'il raconte. nous ne sommes en aucun cas affilé au Grand mistral officiel. Mini-mistral est crée pour rendre hommage au grand Mistral de MistralAI, via une serie de petits modèles écologiques et uniques, se distinguant du gros Mistral, tout en lui rendant hommage.

✨ warning ✨

Mini-Mistral has nothing to do with the Grand Mistral from MistralAI.

It is a completely independent SLM (Single Language Model) from scratch, with its own voice and trained to have its own syntax and its own worldview reflected in the structure of its narrative.

We are in no way affiliated with the official Grand Mistral.

Mini-Mistral was created to pay homage to the Grand Mistral from MistralAI, through a series of small, eco-friendly, and unique models, distinguishing themselves from the larger Mistral while still paying tribute to it.

Downloads last month: 5

Safetensors

Model size

49.1M params

Tensor type

F32

Model tree for Finisha-F-scratch/Mini-mistral-v1

Quantizations

1 model

Space using Finisha-F-scratch/Mini-mistral-v1 1

Collection including Finisha-F-scratch/Mini-mistral-v1

From scratch

Collection

Finisha SLM ✨ • 87 items • Updated Mar 23