Spaces:
Running
Running
update 1
Browse files
app.py
CHANGED
@@ -78,16 +78,14 @@ label_to_class = requests.get(label_map_url).json()
|
|
78 |
# Load your custom model from Hugging Face
|
79 |
model = FineGrainedClassifier(num_classes=len(label_to_class))
|
80 |
model_checkpoint = "Maverick98/EcommerceClassifier"
|
81 |
-
|
|
|
|
|
|
|
|
|
82 |
# Load the tokenizer from Jina
|
83 |
tokenizer = AutoTokenizer.from_pretrained("jinaai/jina-embeddings-v2-base-en")
|
84 |
|
85 |
-
# # Define image preprocessing
|
86 |
-
# transform = transforms.Compose([
|
87 |
-
# transforms.Resize((224, 224)),
|
88 |
-
# transforms.ToTensor(),
|
89 |
-
# transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
|
90 |
-
# ])
|
91 |
|
92 |
def load_image(image_path_or_url):
|
93 |
"""
|
|
|
78 |
# Load your custom model from Hugging Face
|
79 |
model = FineGrainedClassifier(num_classes=len(label_to_class))
|
80 |
model_checkpoint = "Maverick98/EcommerceClassifier"
|
81 |
+
checkpoint_url = f"https://huggingface.co/Maverick98/EcommerceClassifier/resolve/main/model_checkpoint.pth"
|
82 |
+
checkpoint = torch.hub.load_state_dict_from_url(checkpoint_url, map_location=torch.device('cpu'))
|
83 |
+
# Extract and load the model state_dict
|
84 |
+
model.load_state_dict(checkpoint['model_state_dict'])
|
85 |
+
|
86 |
# Load the tokenizer from Jina
|
87 |
tokenizer = AutoTokenizer.from_pretrained("jinaai/jina-embeddings-v2-base-en")
|
88 |
|
|
|
|
|
|
|
|
|
|
|
|
|
89 |
|
90 |
def load_image(image_path_or_url):
|
91 |
"""
|