Update utils/model.py

utils/model.py

@@ -419,10 +419,10 @@ class OwlViTForClassification(nn.Module):
         else:
             #DEUBUG:
             print(f"text_inputs_parts - input_ids: {text_inputs_parts['input_ids'].shape}. attention_mask : {text_inputs_parts['attention_mask'].shape}")
-            seq_length = input_ids.shape[-1]
+            seq_length = text_inputs_parts['input_ids'].shape[-1]
             position_ids = self.owlvit.text_model.embeddings.position_ids[:, :seq_length]
             print(f"position_embedding: {self.owlvit.text_model.embeddings.position_embedding(position_ids).shape}")
-            print(f"text_embeds: {self.owlvit.text_model.text_model.token_embedding(input_ids).shape}")
+            print(f"text_embeds: {self.owlvit.text_model.embeddings.token_embedding(text_inputs_parts['input_ids']).shape}")
             text_embeds_parts = self.owlvit.text_model.text_model.get_text_features(**text_inputs_parts)
             
             # # Embed images and text queries