diff --git a/sqlnet/model/modules/sqlnet_condition_predict.py b/sqlnet/model/modules/sqlnet_condition_predict.py
index 1eb5500..52d5dcd 100644
--- a/sqlnet/model/modules/sqlnet_condition_predict.py
+++ b/sqlnet/model/modules/sqlnet_condition_predict.py
@@ -40,7 +40,7 @@ def __init__(self, N_word, N_h, N_depth, max_col_num, max_tok_num, use_ca, gpu):
         self.cond_col_name_enc = nn.LSTM(input_size=N_word, hidden_size=N_h/2,
                 num_layers=N_depth, batch_first=True,
                 dropout=0.3, bidirectional=True)
-        self.cond_col_out_K = nn.Linear(N_h, N_h)
+        self.cond_col_out_K = nn.Linear(N_h*2, N_h)
         self.cond_col_out_col = nn.Linear(N_h, N_h)
         self.cond_col_out = nn.Sequential(nn.ReLU(), nn.Linear(N_h, 1))
 
@@ -51,7 +51,7 @@ def __init__(self, N_word, N_h, N_depth, max_col_num, max_tok_num, use_ca, gpu):
             self.cond_op_att = nn.Linear(N_h, N_h)
         else:
             self.cond_op_att = nn.Linear(N_h, 1)
-        self.cond_op_out_K = nn.Linear(N_h, N_h)
+        self.cond_op_out_K = nn.Linear(N_h*2, N_h)
         self.cond_op_name_enc = nn.LSTM(input_size=N_word, hidden_size=N_h/2,
                 num_layers=N_depth, batch_first=True,
                 dropout=0.3, bidirectional=True)
@@ -155,6 +155,12 @@ def forward(self, x_emb_var, x_len, col_inp_var, col_name_len,
             col_att = self.softmax(col_att_val.view(
                 (-1, max_x_len))).view(B, -1, max_x_len)
             K_cond_col = (h_col_enc.unsqueeze(1) * col_att.unsqueeze(3)).sum(2)
+
+            # bi-attention
+            temp, _ = torch.max(col_att_val, dim=1)
+            temp_probs = self.softmax(temp).unsqueeze(2)
+            temp2 = (temp_probs*h_col_enc).sum(1)
+            temp2 = temp2.unsqueeze(1).expand([e_cond_col.size()[0],e_cond_col.size()[1],e_cond_col.size()[2]])
         else:
             col_att_val = self.cond_col_att(h_col_enc).squeeze()
             for idx, num in enumerate(x_len):
@@ -164,7 +170,8 @@ def forward(self, x_emb_var, x_len, col_inp_var, col_name_len,
             K_cond_col = (h_col_enc *
                     col_att_val.unsqueeze(2)).sum(1).unsqueeze(1)
 
-        cond_col_score = self.cond_col_out(self.cond_col_out_K(K_cond_col) +
+        cond_col_score = self.cond_col_out(self.cond_col_out_K(
+            torch.cat([K_cond_col,temp2*e_cond_col],dim=-1)) +
                 self.cond_col_out_col(e_cond_col)).squeeze()
         max_col_num = max(col_num)
         for b, num in enumerate(col_num):
@@ -201,6 +208,13 @@ def forward(self, x_emb_var, x_len, col_inp_var, col_name_len,
                     op_att_val[idx, :, num:] = -100
             op_att = self.softmax(op_att_val.view(B*4, -1)).view(B, 4, -1)
             K_cond_op = (h_op_enc.unsqueeze(1) * op_att.unsqueeze(3)).sum(2)
+
+            # bi-attention
+            temp, _ = torch.max(op_att_val, dim=1)
+            temp_probs = self.softmax(temp).unsqueeze(2)
+            temp2 = (temp_probs * h_op_enc).sum(1)
+            temp2 = temp2.unsqueeze(1).expand([col_emb.size()[0], col_emb.size()[1], col_emb.size()[2]])
+
         else:
             op_att_val = self.cond_op_att(h_op_enc).squeeze()
             for idx, num in enumerate(x_len):
@@ -209,8 +223,9 @@ def forward(self, x_emb_var, x_len, col_inp_var, col_name_len,
             op_att = self.softmax(op_att_val)
             K_cond_op = (h_op_enc * op_att.unsqueeze(2)).sum(1).unsqueeze(1)
 
-        cond_op_score = self.cond_op_out(self.cond_op_out_K(K_cond_op) +
-                self.cond_op_out_col(col_emb)).squeeze()
+        cond_op_score = self.cond_op_out(self.cond_op_out_K(
+            torch.cat([K_cond_op,temp2*col_emb],dim=-1)
+        ) + self.cond_op_out_col(col_emb)).squeeze()
 
         #Predict the string of conditions
         h_str_enc, _ = run_lstm(self.cond_str_lstm, x_emb_var, x_len)