Pragyan.ai

🇮🇳 India’s First Open Multilingual AI Model (3B Parameters)

Developed by: Nitin Raj(PragyanXNitinAI )

---

🌟 📌 Model Summary

Pragyan.ai is a powerful 3-billion-parameter multilingual Large Language Model (LLM) created to bring advanced AI capabilities to India and the global community. It is designed for conversation, reasoning, content creation, and coding tasks, with a special focus on Indian languages and cultural context.

🔥 Key Identity: 👉 Pragyan.ai proudly represents the emergence of India’s first AI model vision — lightweight, open, and impactful.

---

🚀 ✨ Key Highlights

• 🇮🇳 India’s First AI Identity crafted for cultural & linguistic relevance.
• 🧠 3B Parameters ensuring high-quality reasoning & generation.
• 🌐 Multilingual Intelligence — English, Hindi, plus partial Indic support.
• ⚡ Optimized for speed and low resource usage (BF16).
• 🔓 Fully Open-Source (Apache 2.0) — Commercial-friendly.
• 🛡 Safety-Aligned with ethical and responsible AI principles.

---

🎯 Intended Use

✔ Recommended Applications

• 🗣️ Conversational AI & Chatbots
• 📝 Content creation — blogs, stories, marketing, etc.
• 🤖 Code generation & developer aid
• 📚 Education & tutoring
• 🔍 Summarization, reasoning & analysis
• 🇮🇳 Indian-language AI systems

🚫 Not Suitable For

• ❌ Medical, legal, or financial decision-making
• ❌ Safety-critical environments
• ❌ Misinformation, harmful, or deceptive activities
• ❌ Fully autonomous decision-making without human review

---

📂 Training Data

Pragyan.ai was trained on a diverse mixture of:

• 📘 Public multilingual datasets (English + Hindi focused)
• 🌍 Open web text
• 🤝 Instruction datasets for alignment & chat abilities

⚠ Note: Training data may contain societal biases, and the model may occasionally reflect them.

---

📈 Model Evaluation

✔ Evaluated On

• Text generation fluency
• Reasoning tasks
• Instruction-following
• English & Hindi conversation

⭐ Performance Observations

• Strong coherence and clarity in English.
• Highly fluent Hindi responses.
• Moderate reasoning capability for a 3B model.
• Occasional hallucinations — verification required.
• Varied accuracy in regional Indic languages.

---

⚠️ Limitations

• May generate incorrect or hallucinated information.
• Limited deep reasoning compared to larger LLMs.
• May misinterpret complex or ambiguous prompts.
• Not designed for critical, expert-level domains.
• Can reflect biases present in training sources.

---

🛡 Ethical Considerations

Users should ensure:

1. ✔ Human supervision for important decisions.
2. ✔ Output filtering where needed.
3. ✔ Avoidance of harmful or malicious use cases.
4. ✔ Transparency when using AI-generated content.

Pragyan.ai is released to empower innovation, not encourage misuse.

---

🧪 How to Use

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "PragyanXNitinAI/Pragyan.ai"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id, 
    torch_dtype=torch.bfloat16, 
    device_map="auto"
)

prompt = "Explain the impact of AI in India."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=200)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))```

📜 License
​📄 Apache License 2.0
You may freely use, modify, distribute, and integrate this model — including in commercial applications — with proper attribution.
​🤝 Community & Contributions
​We welcome:
​⭐ Pull requests
​🔧 Fine-tuned model variants
​🧪 Benchmarks
​🌐 Language expansion contributions
​🐞 Issue reports
​📬 Contact
​Creator: PragyanXNitinAI
​Model Page: https://huggingface.co/PragyanXNitinAI/Pragyan.ai
​✨ Feel free to reach out for collaboration, research, or enterprise integration.
​⭐ If Pragyan.ai inspires you, please star the model — Support open-source AI from India! 🇮🇳