Update requirements and model load

handler.py

@@ -23,6 +23,8 @@ def preprocess_audio(audio_signal, sr, duration=10, target_sr=16000):
 
 class EndpointHandler:
     def __init__(self, path=""):
+        if os.path.isfile("transfer.pth") == False:
+            torch.hub.download_url_to_file('https://huggingface.co/seungheondoh/lp-music-caps/resolve/main/transfer.pth', 'transfer.pth')
         self.device = "cuda:0" if torch.cuda.is_available() else "cpu"
         self.model = BartCaptionModel(max_length = 128)
         pretrained_object = torch.load('./transfer.pth', map_location='cpu')

requirements.txt

@@ -1,19 +1,19 @@
-datasets==2.18.0
-huggingface-hub==0.21.4
-julius==0.2.7
-librosa==0.10.1
+tokenizers==0.13.3
+tqdm==4.66.2
 multidict==6.0.5
 multiprocess==0.70.16
 numpy==1.26.4
 packaging==23.2
 pandas==2.2.1
 pydub==0.25.1
+wavaugment==0.2
 scikit-learn==1.4.1.post1
 scipy==1.12.0
-tokenizers==0.13.3
+librosa==0.10.1
 torch==1.13.1
 torchaudio==0.13.1
 torchaudio-augmentations==0.2.1
-tqdm==4.66.2
 transformers==4.26.1
-wavaugment==0.2
+datasets==2.18.0
+huggingface-hub==0.21.4
+julius==0.2.7