🇮🇳 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:
- ✔ Human supervision for important decisions.
- ✔ Output filtering where needed.
- ✔ Avoidance of harmful or malicious use cases.
- ✔ 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! 🇮🇳
- Downloads last month
- 1