Model Overview

MoALlam-7B-Instruct is an open model instruction-tuned for Tunisian Derja, is a continually pre-trained and aligned version of ALLaM-7B-Instruct-preview with Tunisian_Derja_Dataset

Uploaded model

  • Developed by: Linagora
  • License: apache-2.0
  • Finetuned from model : ALLaM-AI/ALLaM-7B-Instruct-preview

Usage

Below we share some code snippets on how to get quickly started with running the model. First, install the Transformers library with:

pip install transformers

Running with the pipeline API

import torch
from transformers import pipeline
model_id= "linagora/MoALlam-7B-Instruct"
pipe = pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},    
    device="cuda" # replace with "mps" to run on a Mac device
)

messages = [
    {"role": "user", "content": 'وين تجي تونس؟'},
]

outputs = pipe(messages, max_new_tokens=128,  temperature=0.0)
assistant_response = outputs[0]["generated_text"][-1]["content"].strip()
print(assistant_response)
- Response:تونس تجي في شمال إفريقيا بالضبط في الجزء الجنوبي لغربي متاع البحر الأبيض المتوسط.

Running the model on a single / multi GPU

from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto
model_id= "linagora/MoALlam-7B-Instruct"
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="cuda"
)
tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [
    {"role": "user", "content":  "وين تجي تونس؟"},
]

input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt", return_dict=True , add_generation_prompt=True).to(model.device)

outputs = model.generate(**input_ids, max_new_tokens=128)

print(tokenizer.decode(outputs[0]))
- Response:تونس تجي في شمال إفريقيا بالضبط في الجزء الجنوبي لغربي متاع البحر الأبيض المتوسط.

Citations

When using this model ** MoALlam-7B-Instruct **, please cite:

@model{linagora2025LLM-tn,
  author = {Wajdi Ghezaiel and Jean-Pierre Lorré},
  title = {MoALlam-7B-Instruct :Tunisian Arabic Derja LLM based on ALLaM-7B-Instruct},
  year = {2025},
  month = {July},  
  url = {https://huggingface.co/datasets/linagora/MoALlam-7B-Instruct}
}
Downloads last month
21
Safetensors
Model size
7B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for linagora/MoALlam-7B-Instruct

Finetuned
(12)
this model