Instructions to use MohJaf/Bayan-15B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MohJaf/Bayan-15B with Transformers:

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

pipe = pipeline("text-generation", model="MohJaf/Bayan-15B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MohJaf/Bayan-15B")
model = AutoModelForCausalLM.from_pretrained("MohJaf/Bayan-15B")
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 MohJaf/Bayan-15B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MohJaf/Bayan-15B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MohJaf/Bayan-15B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MohJaf/Bayan-15B

SGLang

How to use MohJaf/Bayan-15B 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 "MohJaf/Bayan-15B" \
    --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": "MohJaf/Bayan-15B",
		"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 "MohJaf/Bayan-15B" \
        --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": "MohJaf/Bayan-15B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use MohJaf/Bayan-15B with Docker Model Runner:
```
docker model run hf.co/MohJaf/Bayan-15B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Bayan-15B

A specialized Arabic Large Language Model for legal reasoning, interpretive methodologies, and classical Arabic text analysis.

Model Description

Bayan-15B is a domain-adapted language model built on Qwen2.5-14B, fine-tuned on a comprehensive corpus of classical Arabic legal and interpretive texts. The model excels at understanding complex argumentative structures, legal reasoning patterns, and hermeneutical methodologies in Arabic.

Key Capabilities

Legal Text Analysis: Understanding and generating classical Arabic legal discourse
Interpretive Reasoning: Analyzing methodological frameworks and interpretive principles
Classical Arabic: Deep comprehension of traditional scholarly Arabic writing styles
Argumentation: Following complex chains of reasoning and evidence-based arguments

Training Data

Corpus Size: Approximately 190 million tokens
Sources: Over 900 classical Arabic texts covering legal theory, interpretive methodology, and jurisprudential reasoning
Language: Classical and Modern Standard Arabic

Technical Specifications

Parameter	Value
Base Model	Qwen/Qwen2.5-14B
Parameters	14.7B
Training Method	Continued Pre-Training (CPT)
Context Length	2048 tokens
Precision	bfloat16

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained( "MohJaf/Bayan-15B", torch_dtype="auto", device_map="auto" ) tokenizer = AutoTokenizer.from_pretrained("MohJaf/Bayan-15B")

prompt = "Your Arabic text here" inputs = tokenizer(prompt, return_tensors="pt").to(model.device) outputs = model.generate(**inputs, max_new_tokens=256) print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Use Cases

Academic research in Arabic legal traditions
Analysis of classical interpretive methodologies
Arabic NLP applications requiring domain expertise
Educational tools for Arabic legal studies
Compliance and advisory systems for Islamic finance

Limitations

Specialized in classical Arabic legal discourse
Not a substitute for qualified legal or religious experts
Should be used as a research and analysis tool
May require domain expertise to evaluate outputs

License

This model is released under CC BY-NC-ND 4.0.

Academic and research use is permitted. Commercial use requires separate licensing. Modifications and redistribution are not permitted without prior authorization.

For commercial licensing inquiries, please contact the developer.

Developer

Bayan AI, LLC Building AI solutions for Arabic language understanding and specialized domains.

Citation

@misc {usuli-ai-2025, author = {Bayan AI}, title = {Bayan-15B: Arabic Legal Reasoning Language Model}, year = {2025}, publisher = {Hugging Face}, url = { https://huggingface.co/MohJaf/Bayan-15B } }

Contact

Hugging Face:

@MohJaf

Organization: Bayan AI, LLC

Downloads last month: 15

Safetensors

Model size

15B params

Tensor type

BF16

Model tree for MohJaf/Bayan-15B

Base model

Qwen/Qwen2.5-14B

Finetuned

(107)

this model

Quantizations

2 models