metadata
license: apache-2.0
language:
- en
- sa
- hi
- mr
- gu
- ta
- te
base_model:
- HuggingFaceTB/SmolLM3-3B-Base
🕉️ DeepShiva - AI Travel Companion for Indian Tourism & Wellness
Your intelligent guide to India's spiritual and cultural heritage
🌟 Overview
DeepShiva is a specialized AI model designed to bridge the gap between modern travelers and India's rich spiritual traditions. Built on the robust foundation of SmolLM3-3B-Base, this model serves as your personal companion for exploring Indian tourism, wellness practices, yoga, Ayurveda, and ancient wisdom.
DeepShiva provides culturally-informed, spiritually-aware AI assistance that respects and preserves traditional knowledge while making it accessible to modern practitioners.
🔧 Technical Specifications
- Base Model: SmolLM3-3B-Base (3B parameters)
- Fine-tuning Method: QLoRA (Quantized Low-Rank Adaptation)
- Training Type: Unsupervised Fine-tuning
- Architecture: Transformer-based with specialized Indian cultural knowledge
- Hardware: Trained on AMD MI300 GPU
- Model Size: 3B parameters
🎮 Try the Model
Experience DeepShiva through our interactive web interface:
- Live Demo: Try our Fine-tuned Model
- Hugging Face Space: Available for direct model interaction
- API Access: Available through Hugging Face Inference API
🏃♂️ Quick Start
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
# Load the model and tokenizer
model_name = "Riddhish121/DeepShiva_Indian_Culture"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.float16,
device_map="auto"
)
# Example usage
prompt = "Guide me through a traditional yoga practice for beginners"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=200, temperature=0.7)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)