diff --git a/models/equibind.py b/models/equibind.py
index 72120ae0..4bb94602 100644
--- a/models/equibind.py
+++ b/models/equibind.py
@@ -482,8 +482,8 @@ def forward(self, lig_graph, rec_graph, coords_lig, h_feats_lig, original_ligand
             else:
                 x_evolved_rec = coords_rec
 
-            lig_graph.update_all(fn.copy_edge('msg', 'm'), fn.mean('m', 'aggr_msg'))
-            rec_graph.update_all(fn.copy_edge('msg', 'm'), fn.mean('m', 'aggr_msg'))
+            lig_graph.update_all(fn.copy_e('msg', 'm'), fn.mean('m', 'aggr_msg'))
+            rec_graph.update_all(fn.copy_e('msg', 'm'), fn.mean('m', 'aggr_msg'))
 
             if self.fine_tune:
                 x_evolved_lig = x_evolved_lig + self.att_mlp_cross_coors_V_lig(h_feats_lig) * (
@@ -515,7 +515,7 @@ def forward(self, lig_graph, rec_graph, coords_lig, h_feats_lig, original_ligand
                     Loss = torch.sum((d_squared - geometry_graph.edata['feat'] ** 2)**2) # this is the loss whose gradient we are calculating here
                     grad_d_squared = 2 * (x_evolved_lig[src] - x_evolved_lig[dst])
                     geometry_graph.edata['partial_grads'] = 2 * (d_squared - geometry_graph.edata['feat'] ** 2)[:,None] * grad_d_squared
-                    geometry_graph.update_all(fn.copy_edge('partial_grads', 'partial_grads_msg'),
+                    geometry_graph.update_all(fn.copy_e('partial_grads', 'partial_grads_msg'),
                                               fn.sum('partial_grads_msg', 'grad_x_evolved'))
                     grad_x_evolved = geometry_graph.ndata['grad_x_evolved']
                     x_evolved_lig = x_evolved_lig + self.geometry_reg_step_size * grad_x_evolved