Update app.py

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import yt_dlp as youtube_dl
-from transformers import pipeline, WhisperForConditionalGeneration, WhisperTokenizer
+from transformers import pipeline, BitsAndBytesConfig, WhisperForConditionalGeneration, WhisperTokenizer
 from transformers.pipelines.audio_utils import ffmpeg_read
 import torch
 from huggingface_hub import CommitScheduler
@@ -21,8 +21,17 @@ YT_LENGTH_LIMIT_S = 4800  # 1 hour 20 minutes
 
 device = 0 if torch.cuda.is_available() else "cpu"
 
+
+# Quantization
+bnb_config=BitsAndBytesConfig(
+    load_in_4bit=True
+)
+
 # Load the model
-model = WhisperForConditionalGeneration.from_pretrained(MODEL_NAME)
+model = WhisperForConditionalGeneration.from_pretrained(
+    MODEL_NAME, load_in_4bit=True, device_map="auto"
+)
+
 tokenizer = WhisperTokenizer.from_pretrained(MODEL_NAME)
 
 # Initialize the pipeline with the quantized model