Upload app.py

app.py

@@ -37,32 +37,12 @@ def log_to_ui(message):
 # Initialize embedding model
 embedding_model = SentenceTransformer('all-MiniLM-L6-v2')
 
-# Initialize Llama 3.2 3B model using transformers pipeline with remote inference
-HUGGINGFACE_API_KEY = os.getenv("HUGGINGFACE_API_KEY")
-
-try:
-    if HUGGINGFACE_API_KEY:
-        print("🔐 Logging in to HuggingFace for gated model access...")
-        login(token=HUGGINGFACE_API_KEY)
-        print("✅ Successfully logged in to HuggingFace")
-        
-        print("Initializing Gemma 2 2B via transformers pipeline (remote inference)...")
-        generator = pipeline(
-            "text-generation",
-            model="google/gemma-2-2b-it",
-            token=HUGGINGFACE_API_KEY  # Updated parameter name
-        )
-        print("✅ Gemma 2 2B model initialized successfully")
-        llm_available = True
-    else:
-        print("No HuggingFace API key found - will use template responses")
-        generator = None
-        llm_available = False
-except Exception as e:
-    print(f"Error initializing Gemma 2 2B: {e}")
-    print("Falling back to template responses")
-    generator = None
-    llm_available = False
+# BYOK: No server-side API key initialization
+# All model access will be done via user-provided API keys
+print("🔑 BYOK Mode: No server-side API key configured")
+print("✅ Users will provide their own HuggingFace API keys")
+generator = None
+llm_available = False
 
 # Auto-extract ChromaDB if archive exists and directory is missing/empty
 def setup_chromadb():