anique-1/khaleeji-qwen2.5-7b-finllm

Khaleeji-FinLLM: The First Multi-Lingual Financial LLM for the Gulf Region

Khaleeji-FinLLM is a high-performance, bilingual (Arabic & English) large language model specifically fine-tuned for the financial, legal, and economic landscape of the GCC (Gulf Cooperation Council) countries. Based on the powerful Qwen 2.5 7B Instruct architecture, this model is designed to be the premier agentic expert for the Khaleej.

🌟 Key Highlights

• First of its Kind: The first multi-lingual model dedicated to the specific financial regulations and market dynamics of the Gulf region.
• Bilingual Mastery: Trained on over 500,000 high-quality samples of financial data, expert Q&A, and regulatory documents in both Arabic and English.
• Regional Expertise: Deep understanding of:
  • Saudi Arabia: Zakat calculation (ZATCA), Corporate Tax, Vision 2030 economic shifts.
  • UAE: VAT regulations, Free Zone laws, DIFC/ADGM financial frameworks.
  • Qatar & Kuwait: Energy sector investments and local banking standards.

🚀 Capabilities

Khaleeji-FinLLM excels in tasks requiring specialized regional knowledge:

• Financial Advisory: Complex analysis of Zakat and local taxation.
• Investment Insights: Market analysis for Gulf-based energy, real estate, and fintech sectors.
• Regulatory Compliance: Summarization and interpretation of local financial laws.
• Agentic Reasoning: Capable of acting as a backbone for financial AI agents in the region.

📊 Training Details

• Base Model: Qwen/Qwen2.5-7B-Instruct
• Dataset: 500,000 samples of Khaleeji-specific bilingual financial data.
• Technique: LoRA Fine-tuning (Rank 64) with subsequent full-weight merging.
• Precision: Trained using BFloat16 on AMD MI3110X (ROCm) accelerators.

💻 Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "anique-1/khaleeji-qwen2.5-7b-finllm"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype="auto")

messages = [
    {"role": "system", "content": "You are Khaleeji AI, a premium financial expert for the Gulf region."},
    {"role": "user", "content": "Explain the VAT implications for a retail company in the UAE."}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

generated_ids = model.generate(model_inputs.input_ids, max_new_tokens=512)
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)

📜 License

This model is released under the Apache 2.0 license, in alignment with the base Qwen 2.5 model.

🤝 Citation & Contact

Developed as part of a mission to digitize and democratize financial intelligence in the Gulf. For inquiries or collaboration, please visit the repository or reach out via Hugging Face.