trying cached url again

app.py

@@ -127,6 +127,8 @@ def get_activations(model, image: list, model_name: str,
             break
     output = model(image).detach().cpu().numpy()
     
+    print(model(image), model.model(image))
+    
     image = image.detach().cpu().numpy()
     output_1 = layer_outputs[activation_indices[model_name][0]].detach().cpu().numpy()
     output_2 = layer_outputs[activation_indices[model_name][1]].detach().cpu().numpy()
@@ -271,30 +273,33 @@ def predict_and_analyze(model_name, num_channels, dim, input_channel, image):
     #                                 width_mult=hparams.width_mult,
     #                                 depth_mult=hparams.depth_mult,)
     
-    AutoConfig.register(model_loading_name, EfficientNetConfig)
-    AutoModel.register(EfficientNetConfig, EfficientNetPreTrained)
+    ###### kinda working #####
+    # AutoConfig.register(model_loading_name, EfficientNetConfig)
+    # AutoModel.register(EfficientNetConfig, EfficientNetPreTrained)
+    # model = AutoModel.from_pretrained(model_path + model_loading_name)
+    
     # config = EfficientNetConfig.from_pretrained(model_loading_name)
     
     # model = EfficientNetPreTrained.from_pretrained(model_loading_name)
     # model = AutoModel.from_pretrained(model_loading_name, trust_remote_code=True)
     
-    model = AutoModel.from_pretrained(model_path + model_loading_name)
+    # model = AutoModel.from_pretrained(model_path + model_loading_name)
     
-    # model = EfficientNet(dropout=hparams.dropout,
-    #                                 num_channels=hparams.num_channels,
-    #                                 num_classes=hparams.num_classes,
-    #                                 size=hparams.size,
-    #                                 stochastic_depth_prob=hparams.stochastic_depth_prob,
-    #                                 width_mult=hparams.width_mult,
-    #                                 depth_mult=hparams.depth_mult,)
-    # model_url = cached_download(hf_hub_url(model_loading_name, filename="pytorch_model.bin"))
+    model = EfficientNet(dropout=hparams.dropout,
+                                    num_channels=hparams.num_channels,
+                                    num_classes=hparams.num_classes,
+                                    size=hparams.size,
+                                    stochastic_depth_prob=hparams.stochastic_depth_prob,
+                                    width_mult=hparams.width_mult,
+                                    depth_mult=hparams.depth_mult,)
+    model_url = cached_download(hf_hub_url(model_loading_name, filename="pytorch_model.bin"))
     # print(model_url)
     
-    # loaded = torch.load(model_url, map_location='cpu')
-    # print(loaded)
+    loaded = torch.load(model_url, map_location='cpu')
+    print(loaded)
     
-    # model.load_state_dict(loaded)
-    # print(model)
+    model.load_state_dict(loaded)
+    print(model)
     
     # model = EfficientNetPreTrained(config)
     # config.register_for_auto_class()