syedahafsa58/phase3new

Refactor: Relaxed system prompt to allow general conversation

Update: Production backend logic

Refactor: Enhanced system prompt for better tool calling reliability

Hybrid LLM: Ollama for local, Groq for production

Hardened Ollama: Bake models into image and add diagnostic endpoint

Fix Ollama connection: Switch to 127.0.0.1 and improve startup logging

Production-ready Ollama: Bundle Ollama and pre-pull llama3.2

Fix: Force strip JWT algorithm and address FastAPI deprecations

Fix: strip whitespace from jwt_algorithm to prevent 500 error

Update requirements.txt with stable version constraints

Fix all known dependency conflicts

Fix dependency conflict: upgrade pydantic-settings

Deploy backend to new Space with Docker SDK

initial commit