Update app.py

adjust quantization_config

app.py

@@ -23,8 +23,13 @@ YT_LENGTH_LIMIT_S = 4800  # 1 hour 20 minutes
 device = 0 if torch.cuda.is_available() else "cpu"
 
 # Load model with bitsandbytes quantization
-bnb_config = bnb.QuantizationConfig(bits=4)
-pipe = pipeline(task="automatic-speech-recognition", model=MODEL_NAME, chunk_length_s=30, device=device, quantization_config=bnb_config)
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.float16,
+)
+
+# bnb_config = bnb.QuantizationConfig(bits=4)
+pipe = pipeline(task="automatic-speech-recognition", model=MODEL_NAME, chunk_length_s=30, device=device, quantization_config=quantization_config)
 
 # Define paths and create directory if not exists
 JSON_DATASET_DIR = Path("json_dataset")