update changes

inference_2.py

@@ -96,23 +96,10 @@ get_args(parser)
 args, remaining_args = parser.parse_known_args()
 assert remaining_args == [], remaining_args
 
-# multimodal = load_multimodal_model(args)
 spec_model = load_spec_modality_model(args)
 
-# print(f"Spec model is: {spec_model}")
-
 img_model = load_img_modality_model(args)
 
-# print(f"Image model is: {img_model}")
-
-# spec_in = torch.randn(1, 10_000)
-# rgb_in = torch.randn([1, 3, 256, 256])
-
-# rgb_out = img_model(rgb_in)
-# spec_out = spec_model(spec_in)
-
-# print(f"Img input shape is: {rgb_in.shape}, output shape: {rgb_out}")
-# print(f"Spec input shape is: {spec_in.shape}, output shape is: {spec_out.shape} output: {spec_out}")
 
 def preprocess_img(face):
     face = face / 255
@@ -129,7 +116,7 @@ def deepfakes_spec_predict(input_audio):
     x, _ = input_audio
     audio = preprocess_audio(x)
     spec_grads = spec_model.forward(audio)
-    spec_grads_inv = np.exp(spec_grads.cpu().numpy().squeeze())
+    spec_grads_inv = np.exp(spec_grads.cpu().detach().numpy().squeeze())
 
     # multimodal_grads = multimodal.spec_depth[0].forward(spec_grads)