fix concat tuple

main.py

@@ -5,7 +5,7 @@
     sizes = OptimizationSizes(
         epoch_size=1000,
         train_size=1,
-        eval_size=8,
+        eval_size=16,
         mini_batch_size=1
     )
     model = GravesPredictionNet(nhidden=100)

src/gravesnet.py

@@ -23,7 +23,7 @@ def split_args(m, y, t_x, t_e):
     return gps + t_x, (y_e, t_e)
 
 
-def concat_losses(p, e, t_e):
+def concat_losses(p, e):
     loss_x = -F.sum(F.log(sum_axis(p))) / numpy.float32(p.data.shape[0])
     loss_e = F.sigmoid_cross_entropy(*e)
     return loss_x + loss_e
@@ -32,7 +32,7 @@ def concat_losses(p, e, t_e):
 def loss_func(m, y, t_x, t_e):
     x, e = split_args(m, y, t_x, t_e)
     p = gaussian_mixture_2d_ref(*x)
-    return concat_losses(p, e, t_e)
+    return concat_losses(p, e)
 
 
 # TODO: implement nice gaussian function to plot
@@ -104,9 +104,9 @@ def forward_one_step(self, hidden_state, lstm_cells, x_data, t_x_data, t_e_data,
 
         gps, y_e, hidden_state, lstm_cells = self.bottle_neck(hidden_state, lstm_cells, x_data, train)
         t_x = split_axis_by_widths(t_x, [1, 1])
-        gi, e = gps + t_x, (y_e, t_e)
+        gi, e = (gps + tuple(t_x)), (y_e, t_e)
         p = gaussian_mixture_2d_ref(*gi)
-        loss = concat_losses(p, e, t_e)
+        loss = concat_losses(p, e)
 
         return hidden_state, lstm_cells, loss
 

src/train.py

@@ -69,7 +69,7 @@ def evaluate(context, model, lstm_cells: chainer.Variable,
 
         for t in range(len(es[i]) - 1):
             ci, cx, ce = create_inout(context, x, e, t, mean, stddev)
-            hidden_state, _, loss = model.forward_one_step(hidden_state, lstm_cells, ci, cx, ce, train=False)
+            hidden_state, lstm_cells, loss = model.forward_one_step(hidden_state, lstm_cells, ci, cx, ce, train=False)
             total += loss.data.reshape(())
 
     set_volatile(lstm_cells, False)
@@ -153,7 +153,8 @@ def optimize(model, sizes: OptimizationSizes, data_dir: str):
                     loss_point_train / sizes.eval,
                     loss_seq_train / sizes.eval))
                 sys.stdout.flush()
-                loss_point, loss_seq = evaluate(context, model, lstm_cells, sizes, txs, tes, mean, stddev)
+                lstm_copy = lstm_cells.copy()
+                loss_point, loss_seq = evaluate(context, model, lstm_copy, sizes, txs, tes, mean, stddev)
                 print('\ttest:  [loss/point: {:.6f}, loss/seq: {:.6f}]'.format(loss_point, loss_seq))
                 sys.stdout.flush()
                 loss_point_train = 0.0