utils.py

import asyncio
import ollama

async def generate_stream(query: str):
    """Generates streamed responses from Ollama using LLaMA 3."""
    try:
        stream = ollama.chat(
            model="llama3.2",
            messages=[{"role": "user", "content": query}],
            stream=True
        )

        # Stream output without unnecessary delay
        for chunk in stream:
            if "message" in chunk and "content" in chunk["message"]:
                yield chunk["message"]["content"]  # ✅ No sleep needed

    except Exception as e:
        yield f"⚠️ Error: {str(e)}"