Instructions to use QCRI/Fanar-2-27B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use QCRI/Fanar-2-27B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="QCRI/Fanar-2-27B-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("QCRI/Fanar-2-27B-Instruct")
model = AutoModelForCausalLM.from_pretrained("QCRI/Fanar-2-27B-Instruct")
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 QCRI/Fanar-2-27B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "QCRI/Fanar-2-27B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "QCRI/Fanar-2-27B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/QCRI/Fanar-2-27B-Instruct

SGLang

How to use QCRI/Fanar-2-27B-Instruct 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 "QCRI/Fanar-2-27B-Instruct" \
    --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": "QCRI/Fanar-2-27B-Instruct",
		"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 "QCRI/Fanar-2-27B-Instruct" \
        --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": "QCRI/Fanar-2-27B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use QCRI/Fanar-2-27B-Instruct with Docker Model Runner:
```
docker model run hf.co/QCRI/Fanar-2-27B-Instruct
```

Evaluation

by wahidmounir - opened 2 days ago

Discussion

wahidmounir

2 days ago

I congratulate you for this amazing release
Also I want to point your attention for the Performance Improvements table (Benchmarks)
I've noticed that you have only performed comparaison with the old version performance I think it will be more informative if you also add a performance comparison with the base model (Gemma-3-27B) which will be more informative on the efficiency of the fine tuning process

Hope hear from you
keep the good work

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment