bugfix

app.py

@@ -15,7 +15,7 @@ model = BlipForConditionalGeneration.from_pretrained(
 
 def inference(raw_image, model_n, question, strategy):
     if model_n == 'Image Captioning':
-        image = processor(raw_image).to(device, torch.float16)
+        input = processor(raw_image).to(device, torch.float16)
         with torch.no_grad():
             if strategy == "Beam search":
                 config = GenerationConfig(
@@ -24,7 +24,7 @@ def inference(raw_image, model_n, question, strategy):
                     max_length=20,
                     min_length=5,
                 )
-                captions = model.generate(image, generation_config=config)
+                captions = model.generate(**input, generation_config=config)
             else:
                 config = GenerationConfig(
                     do_sample=True,
@@ -32,7 +32,7 @@ def inference(raw_image, model_n, question, strategy):
                     max_length=20,
                     min_length=5,
                 )
-                captions = model.generate(image, generation_config=config)
+                captions = model.generate(**input, generation_config=config)
             caption = processor.decode(captions[0], skip_special_tokens=True)
             caption = caption.replace(' ', '')
             return 'caption: '+caption