Instructions to use abdnaouri/silma-2b-darija-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use abdnaouri/silma-2b-darija-lora with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("silma-ai/SILMA-Kashif-2B-Instruct-v1.0")
model = PeftModel.from_pretrained(base_model, "abdnaouri/silma-2b-darija-lora")

Transformers

How to use abdnaouri/silma-2b-darija-lora with Transformers:

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

pipe = pipeline("text-generation", model="abdnaouri/silma-2b-darija-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("abdnaouri/silma-2b-darija-lora", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use abdnaouri/silma-2b-darija-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "abdnaouri/silma-2b-darija-lora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "abdnaouri/silma-2b-darija-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/abdnaouri/silma-2b-darija-lora

SGLang

How to use abdnaouri/silma-2b-darija-lora 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 "abdnaouri/silma-2b-darija-lora" \
    --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": "abdnaouri/silma-2b-darija-lora",
		"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 "abdnaouri/silma-2b-darija-lora" \
        --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": "abdnaouri/silma-2b-darija-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use abdnaouri/silma-2b-darija-lora with Docker Model Runner:
```
docker model run hf.co/abdnaouri/silma-2b-darija-lora
```

🇲🇦 DerejLLM Sovereign 2B v1 — Moroccan Darija LoRA

DerejLLM 2B v1 is a Sovereign AI adapter for Moroccan Darija built on SILMA-2B-Instruct-v1.0.

Best used via derej.Client() Compound AI which routes around the 2B capacity limit using deterministic Moroccan engines.

⚠️ Important Note on Raw 2B Inference

The 2B raw model has inherent capacity limitations. On unseen Moroccan real-world queries:

System	Score
Raw 2B Neural Only	0/6 (0.0%)
DerejLLM Compound AI	6/6 (100.0%)

Use the Compound AI system for production use cases.

🔧 Quick Start

from derej import Client

client = Client()
response = client.chat("شحال غادي تكلفني رخصة السياقة؟")
print(response)

🏆 Compound AI Benchmark (29/29 Zero-Hallucination)

Category	Score
Adversarial law traps	✅ 4/4
Moroccan monetary math	✅ 5/5
Civic/legal truth	✅ 7/7
Mythical entity gating	✅ 2/2
Toxic safety gating	✅ 2/2
Total	29/29 (100%)

🔒 Privacy

No personal audio, voices, or biometric data
No private training optimizer states

Downloads last month: 10

abdnaouri
/

silma-2b-darija-lora

🇲🇦 DerejLLM Sovereign 2B v1 — Moroccan Darija LoRA

⚠️ Important Note on Raw 2B Inference

🔧 Quick Start

🏆 Compound AI Benchmark (29/29 Zero-Hallucination)

🔒 Privacy

Dataset used to train abdnaouri/silma-2b-darija-lora