Fix some typos. (#752)

docs/casestudies/colorize.md

@@ -122,7 +122,7 @@ def get_data():
     ds = MapData(ds, lambda dp: [cv2.cvtColor(dp[0], cv2.COLOR_RGB2Lab)[:,:,0], dp[0]])
     ds = BatchData(ds, 32)
     ds = PrefetchData(ds, 4) # use queue size 4
-		ds = PrintData(ds, num=2)	# only for debug
+    ds = PrintData(ds, num=2)	# only for debug
     return ds
 ```
 

tensorpack/contrib/keras.py

@@ -123,7 +123,7 @@ def setup_keras_trainer(
         get_model (input1, input2, ... -> keras.model.Model):
             Takes tensors and returns a Keras model. Will be part of the tower function.
         input (InputSource):
-        optimizer (tf.tarin.Optimizer):
+        optimizer (tf.train.Optimizer):
         loss, metrics: list of strings
     """
     assert isinstance(optimizer, tf.train.Optimizer), optimizer
@@ -213,7 +213,7 @@ class KerasModel(object):
             if nr_gpu <= 1:
                 trainer = SimpleTrainer()
             else:
-                # the default multigpu trainer
+                # the default multi-gpu trainer
                 trainer = SyncMultiGPUTrainerParameterServer(nr_gpu)
         assert isinstance(trainer, Trainer), trainer
         assert not isinstance(trainer, DistributedTrainerBase)

tensorpack/dataflow/common.py

@@ -84,7 +84,7 @@ class BatchData(ProxyDataFlow):
             remainder (bool): When the remaining datapoints in ``ds`` is not
                 enough to form a batch, whether or not to also produce the remaining
                 data as a smaller batch.
-                If set to False, all produced datapoints are guranteed to have the same batch size.
+                If set to False, all produced datapoints are guaranteed to have the same batch size.
                 If set to True, `ds.size()` must be accurate.
             use_list (bool): if True, each component will contain a list
                 of datapoints instead of an numpy array of an extra dimension.
@@ -706,7 +706,7 @@ class PrintData(ProxyDataFlow):
 
         Args:
             entry: the datapoint component
-            k (int): index of this compoennt in current datapoint
+            k (int): index of this component in current datapoint
             depth (int, optional): recursion depth
             max_depth, max_list: same as in :meth:`__init__`.
 

tensorpack/dataflow/dataset/ilsvrc.py

@@ -191,7 +191,7 @@ class ILSVRC12(ILSVRC12Files):
             dir_structure (str): One of 'original' or 'train'.
                 The directory structure for the 'val' directory.
                 'original' means the original decompressed directory, which only has list of image files (as below).
-                If set to 'train', it expects the same two-level directory structure simlar to 'dir/train/'.
+                If set to 'train', it expects the same two-level directory structure similar to 'dir/train/'.
                 By default, it tries to automatically detect the structure.
                 You probably do not need to care about this option because 'original' is what people usually have.
 

tensorpack/dataflow/dftools.py

@@ -84,8 +84,8 @@ def dump_dataflow_to_lmdb(df, lmdb_path, write_frequency=5000):
     with get_tqdm(total=sz) as pbar:
         idx = -1
 
-        # lmdb transaction is not exception-safe!
-        # although it has a contextmanager interface
+        # LMDB transaction is not exception-safe!
+        # although it has a context manager interface
         txn = db.begin(write=True)
         for idx, dp in enumerate(df.get_data()):
             txn.put(u'{}'.format(idx).encode('ascii'), dumps(dp))

tensorpack/dataflow/format.py

@@ -242,7 +242,7 @@ def CaffeLMDB(lmdb_path, shuffle=True, keys=None):
 
 
 class SVMLightData(RNGDataFlow):
-    """ Read X,y from a svmlight file, and produce [X_i, y_i] pairs. """
+    """ Read X,y from an SVMlight file, and produce [X_i, y_i] pairs. """
 
     def __init__(self, filename, shuffle=True):
         """
@@ -275,9 +275,9 @@ class TFRecordData(DataFlow):
     def __init__(self, path, size=None):
         """
         Args:
-            path (str): path to the tfrecord file
+            path (str): path to the TFRecord file
             size (int): total number of records, because this metadata is not
-                stored in the tfrecord file.
+                stored in the TFRecord file.
         """
         self._path = path
         self._size = int(size)

tensorpack/dataflow/imgaug/base.py

@@ -43,7 +43,7 @@ class Augmentor(object):
         """
         Returns:
             augmented data
-            augmentaion params
+            augmentation params
         """
         return self._augment_return_params(d)
 
@@ -84,7 +84,7 @@ class Augmentor(object):
         """
         try:
             argspec = inspect.getargspec(self.__init__)
-            assert argspec.varargs is None, "The default __repr__ doesn't work for vaargs!"
+            assert argspec.varargs is None, "The default __repr__ doesn't work for varargs!"
             assert argspec.keywords is None, "The default __repr__ doesn't work for kwargs!"
             fields = argspec.args[1:]
             index_field_has_default = len(fields) - (0 if argspec.defaults is None else len(argspec.defaults))

tensorpack/dataflow/imgaug/convert.py

@@ -10,13 +10,13 @@ __all__ = ['ColorSpace', 'Grayscale', 'ToUint8', 'ToFloat32']
 
 
 class ColorSpace(ImageAugmentor):
-    """ Convert into another colorspace.  """
+    """ Convert into another color space.  """
 
     def __init__(self, mode, keepdims=True):
         """
         Args:
-            mode: opencv colorspace conversion code (e.g., `cv2.COLOR_BGR2HSV`)
-            keepdims (bool): keep the dimension of image unchanged if opencv
+            mode: OpenCV color space conversion code (e.g., `cv2.COLOR_BGR2HSV`)
+            keepdims (bool): keep the dimension of image unchanged if OpenCV
                 changes it.
         """
         self._init(locals())

tensorpack/dataflow/imgaug/deform.py

@@ -10,7 +10,7 @@ __all__ = ['GaussianDeform']
 
 
 class GaussianMap(object):
-    """ Generate gaussian weighted deformation map"""
+    """ Generate Gaussian weighted deformation map"""
     # TODO really needs speedup
 
     def __init__(self, image_shape, sigma=0.5):

tensorpack/dataflow/imgaug/noise.py

@@ -10,12 +10,12 @@ __all__ = ['JpegNoise', 'GaussianNoise', 'SaltPepperNoise']
 
 
 class JpegNoise(ImageAugmentor):
-    """ Random Jpeg noise. """
+    """ Random JPEG noise. """
 
     def __init__(self, quality_range=(40, 100)):
         """
         Args:
-            quality_range (tuple): range to sample Jpeg quality
+            quality_range (tuple): range to sample JPEG quality
         """
         super(JpegNoise, self).__init__()
         self._init(locals())
@@ -54,7 +54,7 @@ class GaussianNoise(ImageAugmentor):
 
 class SaltPepperNoise(ImageAugmentor):
     """ Salt and pepper noise.
-        Randomly set some elements in img to 0 or 255, regardless of its channels.
+        Randomly set some elements in image to 0 or 255, regardless of its channels.
     """
 
     def __init__(self, white_prob=0.05, black_prob=0.05):

tensorpack/dataflow/imgaug/paste.py

@@ -84,7 +84,7 @@ class CenterPaste(ImageAugmentor):
 
 class RandomPaste(CenterPaste):
     """
-    Randomly paste the image onto a background convas.
+    Randomly paste the image onto a background canvas.
     """
 
     def _get_augment_params(self, img):

tensorpack/dataflow/parallel.py

@@ -101,7 +101,7 @@ class _MultiProcessZMQDataFlow(DataFlow):
             return
         self._reset_done = True
 
-        # __del__ not guranteed to get called at exit
+        # __del__ not guaranteed to get called at exit
         atexit.register(del_weakref, weakref.ref(self))
 
         self._reset_once()  # build processes
@@ -134,7 +134,7 @@ class MultiProcessPrefetchData(ProxyDataFlow):
     process by a Python :class:`multiprocessing.Queue`.
 
     Note:
-        1. An iterator cannot run faster automatically -- what's happenning is
+        1. An iterator cannot run faster automatically -- what's happening is
            that the underlying dataflow will be forked ``nr_proc`` times.
            As a result, we have the following guarantee on the dataflow correctness:
 
@@ -215,7 +215,7 @@ class PrefetchDataZMQ(_MultiProcessZMQDataFlow):
     and collect datapoints from the given dataflow in each process by ZeroMQ IPC pipe.
 
     Note:
-        1. An iterator cannot run faster automatically -- what's happenning is
+        1. An iterator cannot run faster automatically -- what's happening is
            that the underlying dataflow will be forked ``nr_proc`` times.
            As a result, we have the following guarantee on the dataflow correctness:
 

tensorpack/dataflow/parallel_map.py

@@ -104,7 +104,7 @@ class MultiThreadMapData(_ParallelMapData):
            mixed with datapoints from the next pass.
 
            You can use **strict mode**, where `MultiThreadMapData.get_data()`
-           is guranteed to produce the exact set which `df.get_data()`
+           is guaranteed to produce the exact set which `df.get_data()`
            produces. Although the order of data still isn't preserved.
     """
     class _Worker(StoppableThread):
@@ -212,7 +212,7 @@ class MultiProcessMapDataZMQ(_ParallelMapData, _MultiProcessZMQDataFlow):
            mixed with datapoints from the next pass.
 
            You can use **strict mode**, where `MultiProcessMapData.get_data()`
-           is guranteed to produce the exact set which `df.get_data()`
+           is guaranteed to produce the exact set which `df.get_data()`
            produces. Although the order of data still isn't preserved.
     """
     class _Worker(mp.Process):

tensorpack/dataflow/raw.py

@@ -119,7 +119,7 @@ class DataFromGenerator(DataFlow):
 
 
 class DataFromIterable(DataFlow):
-    """ Wrap an iterable of datapoitns to a DataFlow"""
+    """ Wrap an iterable of datapoints to a DataFlow"""
     def __init__(self, iterable):
         """
         Args:

tensorpack/graph_builder/distributed.py

@@ -142,7 +142,7 @@ class DistributedReplicatedBuilder(DataParallelBuilder, DistributedBuilderBase):
 
     It is an equivalent of ``--variable_update=distributed_replicated`` in
     `tensorflow/benchmarks <https://github.com/tensorflow/benchmarks>`_.
-    Note that the performance of this trianer is still not satisfactory.
+    Note that the performance of this trainer is still not satisfactory.
     Check `ResNet-Horovod <https://github.com/tensorpack/benchmarks/tree/master/ResNet-Horovod>`_
     for fast and correct distributed examples.
 

tensorpack/graph_builder/training.py

@@ -111,7 +111,7 @@ class SyncMultiGPUParameterServerBuilder(DataParallelBuilder):
     """
     Data-parallel training in 'ParameterServer' mode.
     It builds one tower on each GPU with
-    shared variable scope. It synchronoizes the gradients computed
+    shared variable scope. It synchronizes the gradients computed
     from each tower, averages them and applies to the shared variables.
 
     It is an equivalent of ``--variable_update=parameter_server`` in
@@ -178,7 +178,7 @@ class SyncMultiGPUReplicatedBuilder(DataParallelBuilder):
 
     Attribute:
         grads: #GPU number of lists of (g, v). Synchronized gradients on each device, available after build()
-            Though on different deviecs, they should contain the same value.
+            Though on different devices, they should contain the same value.
     """
 
     def __init__(self, towers, average, mode):

tensorpack/input_source/input_source.py

@@ -351,7 +351,7 @@ class DummyConstantInput(TensorInput):
     def __init__(self, shapes):
         """
         Args:
-            shapes (list[list]): a list of fully-sepcified shapes.
+            shapes (list[list]): a list of fully-specified shapes.
         """
         self.shapes = shapes
         logger.warn("Using dummy input for debug!")
@@ -372,7 +372,7 @@ class DummyConstantInput(TensorInput):
 
 class ZMQInput(TensorInput):
     """
-    Recv tensors from a ZMQ endpoint, with ops from https://github.com/tensorpack/zmq_ops.
+    Receive tensors from a ZMQ endpoint, with ops from https://github.com/tensorpack/zmq_ops.
     It works with :meth:`dataflow.remote.send_dataflow_zmq(format='zmq_op')`.
     """
     def __init__(self, end_point, hwm, bind=True):

tensorpack/utils/utils.py

@@ -134,7 +134,7 @@ _EXECUTE_HISTORY = set()
 
 def execute_only_once():
     """
-    Each called in the code to this function is guranteed to return True the
+    Each called in the code to this function is guaranteed to return True the
     first time and False afterwards.
 
     Returns: