---
license: mit
base_model: unsloth/meta-llama-3.1-8b-instruct-bnb-4bit
language:
- ar
- en
library_name: transformers
tags:
- lebanese
- ammiya
- arabizi
- nlp
- dgx-spark
- blackwell
- sm_121a
datasets:
- assix/lebanese-dialect-fusion-v1
metrics:
- loss
---

# 🇱🇧 Lebanese-Llama-3.1-8B

**Lebanese-Llama-3.1-8B** is a high-performance LLM fine-tuned specifically for the Lebanese dialect (**Ammiya**). It bridges the gap between Modern Standard Arabic (MSA) and the multi-modal nature of Lebanese communication, seamlessly blending Arabic script, French/English influences, and **Arabizi** (Romanized Arabic with numbers).

This model was trained and validated on the **NVIDIA DGX Spark**, the world’s first personal AI supercomputer powered by the **Grace Blackwell (GB10)** architecture.



## 🎮 Live Demo
Try the model instantly in your browser without any setup:
👉 **[Lebanese Llama Chat Demo](https://huggingface.co/spaces/esix117/lebanese-llama-demo)**


## 🚀 Model Features

* **Dialectal Authenticity:** Fine-tuned to recognize and generate Lebanese syntax, specifically the "B" prefix for verbs and regional idioms (e.g., *mni7*, *shou fi ma fi*).
* **Arabizi Mastery:** Expertly handles Romanized Lebanese using numbers (e.g., `3` for 'ayn, `7` for ha, `2` for hamza).
* **Blackwell Optimized:** Merged into **16-bit (Bfloat16)** to leverage the 5th Gen Tensor Cores and 128GB Unified Memory of the DGX Spark.
* **Cultural Nuance:** Enhanced understanding of Lebanese culinary, geographic, and social context compared to base Llama-3.1.

## 🛠 Training Specifications

* **Infrastructure:** NVIDIA DGX Spark (Grace Blackwell Superchip)
* **Architecture:** Llama-3.1-8B-Instruct (4-bit QLoRA base)
* **Training Steps:** 300 steps
* **Loss Convergence:** Dropped from **5.3** to **2.2285**
* **Optimizer:** Unsloth (Stable Bfloat16 Path)



## 💻 Usage & Implementation

Because this model was developed on the **sm_121a** architecture, it is best loaded using the "Stable Path" to avoid Triton compiler conflicts.

### Standard Inference (Hugging Face Transformers)

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "esix117/lebanese-llama-3.1-8b"
tokenizer = AutoTokenizer.from_pretrained(model_id)

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

# Recommended System Prompt for a Native Persona
system_prompt = "You are a helpful Lebanese assistant speaking strictly in Lebanese Ammiya (dialect)."

messages = [
    {"role": "system", "content": system_prompt},
    {"role": "user", "content": "Marhaba! Kifak el yom? Khabbirni shway shou fi ma fi."}
]

inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to("cuda")
outputs = model.generate(inputs, max_new_tokens=150, temperature=0.7)
# Slice the output to remove the prompt and only show the assistant's reply
print(tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True))
```

⚠️ Known Limitations & Blackwell Optimization

Triton Compatibility: If running on Blackwell hardware (sm_121a), you may encounter a ptxas fatal : Value 'sm_121a' is not defined error when using custom kernels. To fix this, use the standard PyTorch RMS Norm fallback:
```python
import unsloth.kernels.rms_layernorm
unsloth.kernels.rms_layernorm.fast_rms_layernorm = torch.nn.functional.rms_norm
```

🤝 Contribution & Acknowledgements

Developed by esix117 on the DGX Spark infrastructure. Special thanks to the researchers behind the open-source dialect datasets used in this fusion.

License: MIT