Upload app.py

app.py

@@ -6,6 +6,8 @@ import gradio as gr
 
 from fairseq.checkpoint_utils import load_model_ensemble_and_task_from_hf_hub
 from fairseq.models.text_to_speech.hub_interface import TTSHubInterface
+from fairseq.utils import move_to_cuda
+
 
 model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
 feature_extractor = ViTFeatureExtractor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
@@ -31,6 +33,7 @@ gen_kwargs = {"max_length": max_length, "num_beams": num_beams}
 
 
 def inference(image_paths):
+  images = []
   
   #for image_path in image_paths:
   i_image = Image.fromarray(image_paths)
@@ -44,11 +47,16 @@ def inference(image_paths):
 
   preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
   preds = [pred.strip() for pred in preds]
-
+  preds = ' '.join(str(e) for e in preds)
+  #print(preds)
 
   sample = TTSHubInterface.get_model_input(task, preds)
+  sample = move_to_cuda(sample)
+
 
   wav, rate = TTSHubInterface.get_prediction(task, model1, generator, sample)
+  wav = wav.to("cpu")
+  
   return wav