removed BatchKey and fixed dependent parts

Author: devojyotimisra

pvnet/models/multimodal/site_encoders/encoders.py

@@ -74,15 +74,25 @@ def __init__(
         )
 
     def _calculate_attention(self, x):
-        gsp_ids = x["BatchKey.gsp_id"].squeeze().int()
+        i=0
+        for key in x.keys():
+            if(i==17):
+                ans=key
+            print(key, type(key))
+            i+=1
+        gsp_ids = x[ans].squeeze().int()
         attention = self._attention_network(gsp_ids)
         return attention
 
     def _encode_value(self, x):
         # Shape: [batch size, sequence length, PV site]
+        i=0
         for key in x.keys():
-            print(key)
-        pv_site_seqs = x["BatchKey.pv"].float()
+            if(i==6):
+                ans=key
+            print(key, type(key))
+            i+=1
+        pv_site_seqs = x[ans].float()
         batch_size = pv_site_seqs.shape[0]
 
         pv_site_seqs = pv_site_seqs.swapaxes(1, 2).flatten(0, 1)