operations-granite/HuggingFace-Granite-AI-Practice-Agent-Dev

1#!/usr/bin/env python32"""Quick test to check actual Gemini embedding dimensions"""3import os4import requests5 6GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")7EMBEDDING_API_URL = f"https://generativelanguage.googleapis.com/v1beta/models/gemini-embedding-001:embedContent?key={GOOGLE_API_KEY}"8 9print("🔍 Testing Gemini embedding dimensions...")10 11try:12    response = requests.post(13        EMBEDDING_API_URL,14        headers={'Content-Type': 'application/json'},15        json={16            "model": "models/gemini-embedding-001",17            "content": {"parts": [{"text": "test"}]}18        },19        timeout=1020    )21 22    if response.status_code == 200:23        result = response.json()24        embedding = result['embedding']['values']25        print(f"✅ Gemini embedding dimensions: {len(embedding)}")26        print(f"   Model: gemini-embedding-001")27        print(f"   Actual size: {len(embedding)} dimensions")28 29        if len(embedding) == 768:30            print("✅ 768 dimensions - schema is CORRECT")31        elif len(embedding) == 3072:32            print("❌ 3072 dimensions - schema needs UPDATE to vector(3072)!")33        else:34            print(f"⚠️  Unexpected size: {len(embedding)} dimensions")35    else:36        print(f"❌ API error: {response.status_code}")37        print(response.text)38except Exception as e:39    print(f"❌ Error: {e}")40