Update app.py

app.py

@@ -15,9 +15,9 @@ whisper_quant_config = BitsAndBytesConfig(
 
 whisper_model = AutoModelForSpeechSeq2Seq.from_pretrained(
     WHISPER_MODEL,
-    device_map="auto",
-    quantization_config=whisper_quant_config
-)
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,  # Use fp16 if GPU available
+    device_map="auto"
+) 
 
 whisper_tokenizer = AutoTokenizer.from_pretrained(WHISPER_MODEL)
 transcriber = pipeline(
@@ -39,7 +39,11 @@ llama_quant_config = BitsAndBytesConfig(
 
 tokenizer = AutoTokenizer.from_pretrained(LLAMA)
 tokenizer.pad_token = tokenizer.eos_token
-model = AutoModelForCausalLM.from_pretrained(LLAMA, device_map="auto", quantization_config=llama_quant_config)
+model = AutoModelForCausalLM.from_pretrained(
+    LLAMA,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+    device_map="auto"
+)
 
 # Function to Transcribe & Generate Minutes
 def process_audio(audio_file):