Update app.py

app.py

@@ -5,20 +5,16 @@ import gradio as gr
 import platform
 
 device = "cpu" if platform.system() == "Darwin" else "cuda"
+device='cpu'
+MODEL_DIR = "RohanAi/nllb_quantized"
 
-MODEL_DIR = "./nllb-600M-quantized"
-
-# 8-bit quantization for GPU
-bnb_config = BitsAndBytesConfig(load_in_8bit=True)
+# # 8-bit quantization for GPU
+# bnb_config = BitsAndBytesConfig(load_in_8bit=True)
 
 # Load tokenizer + model
 tokenizer = AutoTokenizer.from_pretrained(MODEL_DIR)
-if device == "cuda":
-    model = AutoModelForSeq2SeqLM.from_pretrained(
-        MODEL_DIR, device_map="auto", quantization_config=bnb_config
-    )
-else:
-    model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_DIR)
+
+model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_DIR)
 
 punct_normalizer = MosesPunctNormalizer(lang="en")