Instructions to use Tree-AI-lab/kouman-IA-1b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Tree-AI-lab/kouman-IA-1b with Transformers:

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

pipe = pipeline("text-generation", model="Tree-AI-lab/kouman-IA-1b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Tree-AI-lab/kouman-IA-1b")
model = AutoModelForCausalLM.from_pretrained("Tree-AI-lab/kouman-IA-1b", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Tree-AI-lab/kouman-IA-1b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Tree-AI-lab/kouman-IA-1b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Tree-AI-lab/kouman-IA-1b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Tree-AI-lab/kouman-IA-1b

SGLang

How to use Tree-AI-lab/kouman-IA-1b 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 "Tree-AI-lab/kouman-IA-1b" \
    --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": "Tree-AI-lab/kouman-IA-1b",
		"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 "Tree-AI-lab/kouman-IA-1b" \
        --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": "Tree-AI-lab/kouman-IA-1b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use Tree-AI-lab/kouman-IA-1b with Docker Model Runner:
```
docker model run hf.co/Tree-AI-lab/kouman-IA-1b
```

Kouman - Modèle Baoulé / Français (1B)

Kouman est un modèle de langage (basé sur Llama-3.2-1B) pré-entraîné et finetuné spécifiquement pour la traduction et la conversation en Baoulé (bci) et en Français (fr).

Ce dépôt contient le modèle complet fusionné au format safetensors, prêt à l'emploi.

Comment l'utiliser (Code d'inférence)

Voici comment charger et utiliser le modèle avec la bibliothèque transformers de Hugging Face :

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

repo_id = "Tree-AI-lab/kouman-IA-1b"

# Chargement du tokenizer et du modèle
tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = AutoModelForCausalLM.from_pretrained(
    repo_id,
    torch_dtype=torch.float16,
    device_map="auto"
)

# Préparation du prompt
messages = [
    {"role": "system", "content": "Tu es Kouman, un assistant IA multilingue."},
    {"role": "user", "content": "Traduis en baoulé : Bonjour, comment vas-tu aujourd'hui ?"}
]

prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

# Génération de la réponse
with torch.no_grad():
    outputs = model.generate(**inputs, max_new_tokens=150, temperature=0.7)

# Affichage du résultat
reponse = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
print(reponse)

Downloads last month: 27

Safetensors

Model size

1B params

Tensor type

BF16