Update modeling_paragram_sp.py

modeling_paragram_sp.py

@@ -28,7 +28,7 @@ class ParagramSPModel(BertPreTrainedModel):
         print(input_ids)
         print(attention_mask)
         input_ids = self.filter_input_ids(input_ids)
-        attention_mask = input_ids > 0
+        attention_mask = input_ids != self.config.pad_token_id
         embeddings = self.word_embeddings(input_ids)
         masked_embeddings = embeddings * attention_mask[:, :, None]
         mean_pooled_embeddings = masked_embeddings.sum(dim=1) / attention_mask[:, :, None].sum(dim=1)