Update app.py

app.py

@@ -7,12 +7,10 @@ os.environ['CURL_CA_BUNDLE'] = ''
 
 device = 'cuda' if torch.cuda.is_available() else 'cpu'
 
-config = LinearMappingConfig(image_from_pretrained=False, text_from_pretrained=False)
-model = LinearMapping(config)
+config = CLIPGPT2Config(image_from_pretrained=False, text_from_pretrained=False)
+model = CLIPGPT2(config)
 model.load_state_dict(torch.load("pytorch_model.bin", map_location=device))
-processor = LinearMappingProcessor(config)
-processor.tokenizer.padding_side = 'left'
-processor.tokenizer.pad_token_id = processor.tokenizer.eos_token_id
+processor = CLIPGPT2Processor(config)
 
 title = "Generate Image Captions With CLIP And GPT2"
 
@@ -28,11 +26,12 @@ def generate_image_captions(image, text):
         attention_mask=attention_mask,
         max_new_tokens=50
     )
-
+    processor.tokenizer.padding_side = 'left'
+    processor.tokenizer.pad_token_id = processor.tokenizer.eos_token_id
     prediction_text = processor.decode(prediction[0], skip_special_tokens=True)
+    
     return prediction_text
 
-
 article = "This demo is originated from this paper: [original paper](https://arxiv.org/abs/2209.15162)"
 description = """
 ### Expand GPT2's language capabilities to vision with CLIP!