Instructions to use ViraIntelligentDataMining/PersianLLaMA-13B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ViraIntelligentDataMining/PersianLLaMA-13B with Transformers:

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

pipe = pipeline("text-generation", model="ViraIntelligentDataMining/PersianLLaMA-13B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ViraIntelligentDataMining/PersianLLaMA-13B")
model = AutoModelForCausalLM.from_pretrained("ViraIntelligentDataMining/PersianLLaMA-13B")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ViraIntelligentDataMining/PersianLLaMA-13B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ViraIntelligentDataMining/PersianLLaMA-13B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ViraIntelligentDataMining/PersianLLaMA-13B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/ViraIntelligentDataMining/PersianLLaMA-13B

SGLang

How to use ViraIntelligentDataMining/PersianLLaMA-13B 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 "ViraIntelligentDataMining/PersianLLaMA-13B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ViraIntelligentDataMining/PersianLLaMA-13B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "ViraIntelligentDataMining/PersianLLaMA-13B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ViraIntelligentDataMining/PersianLLaMA-13B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use ViraIntelligentDataMining/PersianLLaMA-13B with Docker Model Runner:
```
docker model run hf.co/ViraIntelligentDataMining/PersianLLaMA-13B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

PersianLLaMA: Towards Building First Persian Large Language Model

🌟 Introduction

Welcome to the home of PersianLLaMA, the pioneering large language model for the Persian language. With 13 billion parameters, this model is trained on Persian Wikipedia corpus and designed to excel in multiple NLP tasks, setting a new benchmark for Persian language understanding and generation.

🛠 Model Description

PersianLLaMA is not just a model but a comprehensive tool for:

📝 Text Generation: Crafting coherent and contextually appropriate text.
🎯 Instruct Tuning: Executing tasks based on detailed instructions, ideal for scenarios where the model needs to adhere to specific guidelines or produce outputs tailored to particular requirements.
❓ Question Answering: Providing accurate answers to Persian queries.
📊 Text Summarization: Condensing Persian texts into precise summaries.

This model has been collaboratively developed by a team of experts, including Mohammad Amin Abbasi, Arash Ghafouri, Mahdi Firouzmandi, Hassan Naderi, Behrouz Minaei Bidgoli.

🚀 Quick Start

To integrate PersianLLaMA into your project, follow these steps:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "ViraIntelligentDataMining/PersianLLaMA-13B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

prompt = "این متن به فارسی است"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(inputs["input_ids"])
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

📈 Evaluation and Benchmarks

PersianLLaMA demonstrates superior performance over existing models, with robust evaluation metrics that highlight its capabilities in natural language understanding and generation.

📜 Citing PersianLLaMA

If you find PersianLLaMA useful in your research, please consider citing:

@article{abbasi2023persianllama,
  title={PersianLLaMA: Towards Building First Persian Large Language Model},
  author={Abbasi, Mohammad Amin and others},
  journal={https://arxiv.org/abs/2312.15713},
  year={2023}
}