saving stuff

Author: StephanieLarocque

code/cnn_1D_segm/train_fcn1D.py

@@ -337,24 +337,18 @@ def jaccard(y_pred, y_true, n_classes, one_hot=False):
         print('saving best (and last) model')
         best_jacc_val = jacc_valid[epoch]
         patience = 0
-        np.savez(os.path.join(savepath, 'new_fcn1D_model_best.npz'),
-                 *lasagne.layers.get_all_param_values(simple_net_output))
+        np.savez(os.path.join(savepath, 'new_fcn1D_model_best.npz'),  *lasagne.layers.get_all_param_values(simple_net_output))
         np.savez(os.path.join(savepath , "fcn1D_errors_best.npz"),
                  err_train=err_train, acc_train=acc_train,
                  err_valid=err_valid, acc_valid=acc_valid, jacc_valid=jacc_valid)
-        np.savez(os.path.join(savepath, 'new_fcn1D_model_last.npz'),
-                 *lasagne.layers.get_all_param_values(simple_net_output))
-        np.savez(os.path.join(savepath , "fcn1D_errors_last.npz"),
-                 err_train=err_train, acc_train=acc_train,
-                 err_valid=err_valid, acc_valid=acc_valid, jacc_valid=jacc_valid)
     else:
         patience += 1
         print('saving last model')
-        np.savez(os.path.join(savepath, 'new_fcn1D_model_last.npz'),
-                 *lasagne.layers.get_all_param_values(simple_net_output))
-        np.savez(os.path.join(savepath , "fcn1D_errors_last.npz"),
-                 err_train=err_train, acc_train=acc_train,
-                 err_valid=err_valid, acc_valid=acc_valid, jacc_valid=jacc_valid)
+
+    np.savez(os.path.join(savepath, 'new_fcn1D_model_last.npz'),  *lasagne.layers.get_all_param_values(simple_net_output))
+    np.savez(os.path.join(savepath , "fcn1D_errors_last.npz"),
+             err_train=err_train, acc_train=acc_train,
+             err_valid=err_valid, acc_valid=acc_valid, jacc_valid=jacc_valid)
     # Finish training if patience has expired or max nber of epochs reached
 
     if patience == max_patience or epoch == num_epochs-1:

code/fcn_2D_segm/train_fcn8.py

@@ -213,6 +213,10 @@ def train(dataset, learn_step=0.005,
     jacc_valid = []
     patience = 0
 
+    n_batches_train = 1
+    n_batches_val = 1
+    n_batches_test = 1
+    num_epochs = 1
     # Training main loop
     print "Start training"
     for epoch in range(num_epochs):
@@ -222,10 +226,12 @@ def train(dataset, learn_step=0.005,
 
         # Train
         for i in range(n_batches_train):
+            print 'Training batch ', i
             # Get minibatch
             X_train_batch, L_train_batch = train_iter.next()
             L_train_batch = np.reshape(L_train_batch, np.prod(L_train_batch.shape))
 
+
             # Training step
             cost_train = train_fn(X_train_batch, L_train_batch)
             out_str = "cost %f" % (cost_train)
@@ -238,6 +244,7 @@ def train(dataset, learn_step=0.005,
         acc_val_tot = 0
         jacc_val_tot = np.zeros((2, n_classes))
         for i in range(n_batches_val):
+            print 'Valid batch ', i
             # Get minibatch
             X_val_batch, L_val_batch = val_iter.next()
             L_val_batch = np.reshape(L_val_batch, np.prod(L_val_batch.shape))
@@ -277,13 +284,12 @@ def train(dataset, learn_step=0.005,
             best_jacc_val = jacc_valid[epoch]
             patience = 0
             np.savez(os.path.join(savepath, 'new_fcn8_model_best.npz'),  *lasagne.layers.get_all_param_values(convmodel))
-            np.savez(os.path.join(savepath + "fcn8_errors_best.npz"),
-                     err_valid, err_train, acc_valid, jacc_valid)
+            np.savez(os.path.join(savepath, "fcn8_errors_best.npz"),  err_valid, err_train, acc_valid, jacc_valid)
         else:
             patience += 1
-            np.savez(os.path.join(savepath, 'new_fcn8_model_last.npz'), *lasagne.layers.get_all_param_values(convmodel))
-            np.savez(os.path.join(savepath + "fcn8_errors_last.npz"),
-                     err_valid, err_train, acc_valid, jacc_valid)
+
+        np.savez(os.path.join(savepath, 'new_fcn8_model_last.npz'), *lasagne.layers.get_all_param_values(convmodel))
+        np.savez(os.path.join(savepath, "fcn8_errors_last.npz"),  err_valid, err_train, acc_valid, jacc_valid)
         # Finish training if patience has expired or max nber of epochs
         # reached
         if patience == max_patience or epoch == num_epochs-1:

code/unet/train_unet.py

@@ -226,9 +226,6 @@ def train(dataset, learn_step=0.005,
         # Single epoch training and validation
         start_time = time.time()
         cost_train_tot = 0
-
-        n_batches_train = 2
-        n_batches_val = 2
         # Train
         print 'Training steps '
         for i in range(n_batches_train):
@@ -291,13 +288,12 @@ def train(dataset, learn_step=0.005,
             best_jacc_val = jacc_valid[epoch]
             patience = 0
             np.savez(os.path.join(savepath, 'new_unet_model_best.npz'),  *lasagne.layers.get_all_param_values(output_layer))
-            np.savez(os.path.join(savepath + "unet_errors_best.npz"),
-                     err_valid, err_train, acc_valid, jacc_valid)
+            np.savez(os.path.join(savepath, 'unet_errors_best.npz'), err_valid, err_train, acc_valid, jacc_valid)
         else:
             patience += 1
-            np.savez(os.path.join(savepath, 'new_unet_model_last.npz'), *lasagne.layers.get_all_param_values(output_layer))
-            np.savez(os.path.join(savepath + "unet_errors_last.npz"),
-                     err_valid, err_train, acc_valid, jacc_valid)
+            
+        np.savez(os.path.join(savepath, 'new_unet_model_last.npz'), *lasagne.layers.get_all_param_values(output_layer))
+        np.savez(os.path.join(savepath, 'unet_errors_last.npz'),  err_valid, err_train, acc_valid, jacc_valid)
         # Finish training if patience has expired or max nber of epochs
         # reached
         if patience == max_patience or epoch == num_epochs-1: