Sahjid123/safarnama-gemma2-light

SafarNama.AI - Gemma 2B Lightweight Model

Fine-tuned Gemma-2-2B-IT for Pakistan travel queries (lightweight model).

Model Details

• Base Model: google/gemma-2-2b-it
• Fine-tuned for: Pakistan travel Q&A
• Task: Quick factual questions about destinations, hotels, attractions
• Size: ~2 GB (4-bit quantized)
• Response Time: 2-5 seconds

Use Cases

• ✅ "What is Faisal Mosque?"
• ✅ "Tell me about Badshahi Mosque"
• ✅ "Where is Hunza Valley?"
• ✅ "What's the rating of Hotel XYZ?"

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Sahjid123/safarnama-gemma2-light"

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

messages = [{"role": "user", "content": "What is Faisal Mosque?"}]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=150)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

Part of SafarNama.AI

This is the lightweight model in the SafarNama.AI dual-model system:

• Lightweight (this model): Fast Q&A
• Heavy: Detailed itinerary planning

License

Apache 2.0