refactor(core): Switch to TinyLlama to resolve memory issues

api/main.py

@@ -45,8 +45,8 @@ ROOT_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 MODELS_DIR = os.path.join(ROOT_DIR, 'models') 
 
 # ✅ FIX: Swapped to a smaller, memory-friendly model to avoid crashing on free tier
-MODEL_REPO = "TheBloke/phi-2-GGUF"
-MODEL_FILENAME = "phi-2.Q2_K.gguf" 
+MODEL_REPO = "TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF"
+MODEL_FILENAME = "tinyllama-1.1b-chat-v1.0.Q2_K.gguf" 
 
 MODEL_SAVE_DIRECTORY = os.path.join(os.environ.get("WRITABLE_DIR", "/data"), "llm_model")
 LLAMA_MODEL_PATH = os.path.join(MODEL_SAVE_DIRECTORY, MODEL_FILENAME)