islamic-ai-foundation

🌙 Islamic AI Foundation Model

World's first foundation model for comprehensive Islamic knowledge processing with Noor-e-Abjad integration

🧠 Enhanced Capabilities

• Quranic Analysis: Recitation, Tajweed correction, Abjad validation with Jannah Points
• Hadith Processing: Authentication, classification, isnad analysis with Sahih collections
• Fiqh Understanding: Multi-madhhab rulings, contemporary issues with Islamic QA
• Multilingual Support: Arabic, Urdu, English with proper diacritics handling
• Ethical Compliance: Quranic principles built-in with ADANiD proprietary license
• Noor-e-Abjad Integration: Classical Hisab al-Jummal with 786 validation

🔧 Universal Integration

As Base Model

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("adnanmd76/islamic-ai-foundation")
tokenizer = AutoTokenizer.from_pretrained("adnanmd76/islamic-ai-foundation")

For Fine-tuning with Noor-e-Abjad Dataset

from datasets import load_dataset

# Load Noor-e-Abjad dataset
dataset = load_dataset("adnanmd76/nooreabjad-dataset")
# Your fine-tuning code herefrom transformers import Trainer, TrainingArguments

training_args = TrainingArguments(
    output_dir="./nooreabjad-finetuned",
    per_device_train_batch_size=4,
    num_train_epochs=3,
    fp16=True
)

trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=dataset["train"]
)

For Merging with Other Models

# mergekit-multimodal-islamic.yaml
models:
  - model: adnanmd76/islamic-ai-foundation
    weight: 0.6
    parameters:
      - name: classifier
        weight: 0.7
  - model: aubmindlab/bert-base-arabertv2
    weight: 0.4
    parameters:
      - name: encoder
        weight: 0.3
merge_method: linear
dtype: float16

API Integration

# FastAPI integration
from fastapi import FastAPI
app = FastAPI()

@app.post("/validate-abjad")
async def validate_abjad(text: str):
    inputs = tokenizer(text, return_tensors="pt")
    outputs = model(**inputs)
    return {"prediction": outputs.logits.argmax().item()}

🔗 Ecosystem Integration

• Kaggle Dataset: nooreabjad-dataset
• GitHub Repository: adan-id-opencloud- Hugging Face Dataset: Quranlab-islamic-dataset
• Live Demo: Quranlab Demo Space

🔒 ADANiD Proprietary License

• ✅ FREE for educational, non-commercial use
• ✅ PERMITTED: Madaris, universities, non-profit Islamic organizations
• ✅ ALLOWED: Personal spiritual development and Quranic recitation practice
• ❌ PROHIBITED: Commercial applications without paid license
• ❌ ABSOLUTELY FORBIDDEN: Military, government, or enterprise use

"And We have certainly made the Qur’an easy for remembrance..." — Quran 54:17 "Read in the name of your Lord who created..." — Quran 96:1