Update app.py

app.py

@@ -3,17 +3,14 @@ import re
 import gradio as gr
 from transformers import AutoTokenizer, ViTFeatureExtractor, VisionEncoderDecoderModel 
 
-    
-device = 'cpu'
-encoder_checkpoint = 'google/vit-base-patch16-224-in21k'
-decoder_checkpoint = 'distilgpt2'
-model_checkpoint = '"gagan3012/ViTGPT2_vizwiz"'
+encoder_checkpoint = "google/vit-base-patch16-224-in21k"
+decoder_checkpoint = "gpt2"
+model_checkpoint = "gagan3012/ViTGPT2I2A"
 feature_extractor = ViTFeatureExtractor.from_pretrained(encoder_checkpoint)
 tokenizer = AutoTokenizer.from_pretrained(decoder_checkpoint)
 model = VisionEncoderDecoderModel.from_pretrained(model_checkpoint).to(device)
 
 
-
 def predict(image,max_length=64, num_beams=4):
   image = image.convert('RGB')
   image = feature_extractor(image, return_tensors="pt").pixel_values.to(device)