Update app.py

app.py

@@ -12,7 +12,7 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 import logging
 
 # Configure logging
-logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
+logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %(message)s')
 
 # --- Authentication and Configuration ---  (Moved BEFORE model loading)
 try:
@@ -62,7 +62,7 @@ def transcribe_audio(audio_path: str, whisper_model) -> str:  # Pass whisper_mod
 
 def generate_response(text: str, model_gemma, tokenizer_gemma, device) -> str: # Pass model and tokenizer
     try:
-        input_text = "Here is a response for the user. " + text
+        input_text = "Here is a response to the user: " + text
         input = tokenizer_gemma(input_text, return_tensors="pt").to(device)
         generated_output = model_gemma.generate(**input, max_length=MAX_GEMMA_LENGTH, early_stopping=True)
         return tokenizer_gemma.decode(generated_output[0], skip_special_tokens=True)
@@ -111,7 +111,7 @@ def infer(user_audio) -> tuple[int, np.ndarray]:
         logging.info("Whisper model loaded successfully.")
 
         tokenizer_gemma = AutoTokenizer.from_pretrained("google/gemma-3-1b-pt")
-        model_gemma = AutoModelForCausalLM.from_pretrained("google/gemma-3-1b-pt").to(device)
+        model_gemma = AutoModelForCausalLM.from_pretrained("google/gemma-3-1b-it").to(device)
         logging.info("Gemma 3 1B pt model loaded successfully.")
 
         if not user_audio: