bnrelu, classificationerror

examples/cifar10_convnet.py

@@ -20,8 +20,6 @@
 
 """
 CIFAR10 90% validation accuracy after 70k step.
-
-91% validation accuracy after 36k step with 3 GPU.
 """
 
 BATCH_SIZE = 128
@@ -46,15 +44,15 @@ def _get_cost(self, input_vars, is_training):
 
         image = image / 4.0     # just to make range smaller
         l = Conv2D('conv1.1', image, out_channel=64, kernel_shape=3)
-        l = Conv2D('conv1.2', l, out_channel=64, kernel_shape=3, nl=BNReLU(is_training))
+        l = Conv2D('conv1.2', l, out_channel=64, kernel_shape=3, nl=BNReLU(is_training), use_bias=False)
         l = MaxPooling('pool1', l, 3, stride=2, padding='SAME')
 
         l = Conv2D('conv2.1', l, out_channel=128, kernel_shape=3)
-        l = Conv2D('conv2.2', l, out_channel=128, kernel_shape=3, nl=BNReLU(is_training))
+        l = Conv2D('conv2.2', l, out_channel=128, kernel_shape=3, nl=BNReLU(is_training), use_bias=False)
         l = MaxPooling('pool2', l, 3, stride=2, padding='SAME')
 
         l = Conv2D('conv3.1', l, out_channel=128, kernel_shape=3, padding='VALID')
-        l = Conv2D('conv3.2', l, out_channel=128, kernel_shape=3, padding='VALID', nl=BNReLU(is_training))
+        l = Conv2D('conv3.2', l, out_channel=128, kernel_shape=3, padding='VALID', nl=BNReLU(is_training), use_bias=False)
         l = FullyConnected('fc0', l, 1024 + 512,
                            b_init=tf.constant_initializer(0.1))
         l = tf.nn.dropout(l, keep_prob)
@@ -69,7 +67,7 @@ def _get_cost(self, input_vars, is_training):
         cost = tf.reduce_mean(cost, name='cross_entropy_loss')
         tf.add_to_collection(MOVING_SUMMARY_VARS_KEY, cost)
 
-        # compute the number of failed samples, for ValidationError to use at test time
+        # compute the number of failed samples, for ClassificationError to use at test time
         wrong = prediction_incorrect(logits, label)
         nr_wrong = tf.reduce_sum(wrong, name='wrong')
         # monitor training error
@@ -125,7 +123,7 @@ def get_config():
     lr = tf.train.exponential_decay(
         learning_rate=1e-2,
         global_step=get_global_step_var(),
-        decay_steps=dataset_train.size() * 30 if nr_gpu == 1 else 15,
+        decay_steps=dataset_train.size() * 30 if nr_gpu == 1 else 20,
         decay_rate=0.5, staircase=True, name='learning_rate')
     tf.scalar_summary('learning_rate', lr)
 
@@ -135,7 +133,7 @@ def get_config():
         callbacks=Callbacks([
             StatPrinter(),
             PeriodicSaver(),
-            ValidationError(dataset_test, prefix='test'),
+            ClassificationError(dataset_test, prefix='test'),
         ]),
         session_config=sess_config,
         model=Model(),
@@ -155,6 +153,8 @@ def get_config():
 
     if args.gpu:
         os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
+    else:
+        os.environ['CUDA_VISIBLE_DEVICES'] = '0'
 
     with tf.Graph().as_default():
         config = get_config()

examples/cifar10_resnet.py

@@ -167,7 +167,7 @@ def get_config():
         callbacks=Callbacks([
             StatPrinter(),
             PeriodicSaver(),
-            ValidationError(dataset_test, prefix='test'),
+            ClassificationError(dataset_test, prefix='test'),
             ScheduledHyperParamSetter('learning_rate',
                                       [(1, 0.1), (82, 0.01), (123, 0.001), (300, 0.0002)])
         ]),

examples/mnist_convnet.py

@@ -62,7 +62,7 @@ def _get_cost(self, input_vars, is_training):
         cost = tf.reduce_mean(cost, name='cross_entropy_loss')
         tf.add_to_collection(MOVING_SUMMARY_VARS_KEY, cost)
 
-        # compute the number of failed samples, for ValidationError to use at test time
+        # compute the number of failed samples, for ClassificationError to use at test time
         wrong = prediction_incorrect(logits, label)
         nr_wrong = tf.reduce_sum(wrong, name='wrong')
         # monitor training error
@@ -106,7 +106,7 @@ def get_config():
             StatPrinter(),
             PeriodicSaver(),
             ValidationStatPrinter(dataset_test, ['cost:0']),
-            ValidationError(dataset_test, prefix='validation'),
+            ClassificationError(dataset_test, prefix='validation'),
         ]),
         session_config=sess_config,
         model=Model(),

examples/svhn_digit_convnet.py

@@ -53,7 +53,7 @@ def _get_cost(self, input_vars, is_training):
         cost = tf.reduce_mean(cost, name='cross_entropy_loss')
         tf.add_to_collection(MOVING_SUMMARY_VARS_KEY, cost)
 
-        # compute the number of failed samples, for ValidationError to use at test time
+        # compute the number of failed samples, for ClassificationError to use at test time
         wrong = prediction_incorrect(logits, label)
         nr_wrong = tf.reduce_sum(wrong, name='wrong')
         # monitor training error
@@ -110,7 +110,7 @@ def get_config():
         callbacks=Callbacks([
             StatPrinter(),
             PeriodicSaver(),
-            ValidationError(test, prefix='test'),
+            ClassificationError(test, prefix='test'),
         ]),
         session_config=sess_config,
         model=Model(),

tensorpack/callbacks/validation_callback.py

@@ -12,7 +12,7 @@
 from ..tfutils.summary import *
 from .base import PeriodicCallback, Callback, TestCallbackType
 
-__all__ = ['ValidationError', 'ValidationCallback', 'ValidationStatPrinter']
+__all__ = ['ClassificationError', 'ValidationCallback', 'ValidationStatPrinter']
 
 class ValidationCallback(PeriodicCallback):
     """
@@ -100,8 +100,7 @@ def _trigger_periodic(self):
                 '{}_{}'.format(self.prefix, name), stat), self.global_step)
             self.trainer.stat_holder.add_stat("{}_{}".format(self.prefix, name), stat)
 
-
-class ValidationError(ValidationCallback):
+class ClassificationError(ValidationCallback):
     """
     Validate the accuracy from a `wrong` variable
 
@@ -119,7 +118,7 @@ def __init__(self, ds, prefix='validation',
         :param ds: a batched `DataFlow` instance
         :param wrong_var_name: name of the `wrong` variable
         """
-        super(ValidationError, self).__init__(ds, prefix, period)
+        super(ClassificationError, self).__init__(ds, prefix, period)
         self.wrong_var_name = wrong_var_name
 
     def _find_output_vars(self):