2 hf

README.md

@@ -9,6 +9,4 @@ app_file: app.py
 pinned: false
 license: mit
 short_description: Chinese Traditional Instrument Sound Retriever
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+---

model.py

@@ -1,7 +1,7 @@
 import torch
 import torch.nn as nn
 import torchvision.models as models
-from modelscope.msdatasets import MsDataset
+from datasets import load_dataset
 from utils import MODEL_DIR
 
 
@@ -38,11 +38,7 @@ class EvalNet:
         return backbone_list[0]
 
     def _model_info(self, m_ver: str):
-        backbone_list = MsDataset.load(
-            "monetjoe/cv_backbones",
-            split="v1",
-            trust_remote_code=True,
-        )
+        backbone_list = load_dataset("monetjoe/cv_backbones", split="train")
         backbone = self._get_backbone(m_ver, backbone_list)
         m_type = str(backbone["type"])
         input_size = int(backbone["input_size"])

requirements.txt

@@ -2,5 +2,4 @@ torch
 pillow
 librosa
 matplotlib
-torchvision
-modelscope==1.15
+torchvision

utils.py

@@ -4,10 +4,7 @@ import torchvision.transforms as transforms
 from huggingface_hub import snapshot_download
 from PIL import Image
 
-MODEL_DIR = snapshot_download(
-    "ccmusic-database/CTIS",
-    cache_dir="./__pycache__",
-)
+MODEL_DIR = snapshot_download("ccmusic-database/CTIS", cache_dir="./__pycache__")
 
 
 def toCUDA(x):