update docs

.lgtm.yml

@@ -13,4 +13,5 @@ extraction:
     prepare:
       packages:
         - libcap-dev
-
+    python_setup:
+      version: 3

docs/conf.py

@@ -83,7 +83,10 @@ if ON_RTD:
 else:
     # skip this when building locally
     intersphinx_timeout = 0.1
-intersphinx_mapping = {'python': ('https://docs.python.org/3.6', None)}
+intersphinx_mapping = {
+    'python': ('https://docs.python.org/3.6', None),
+    'numpy': ('https://docs.scipy.org/doc/numpy/', None),
+}
 # -------------------------
 
 # Add any paths that contain templates here, relative to this directory.
@@ -106,7 +109,7 @@ master_doc = 'index'
 
 # General information about the project.
 project = u'tensorpack'
-copyright = u'2015 - 2018, Yuxin Wu, et al.'
+copyright = u'2015 - 2019, Yuxin Wu, et al.'
 author = u'Yuxin Wu, et al.'
 
 # The version info for the project you're documenting, acts as replacement for

docs/tutorial/dataflow.md

@@ -55,7 +55,7 @@ Mappers execute a mapping function in parallel on top of an existing dataflow.
 You can find details in the [API docs](../modules/dataflow.html) under the
 "parallel" and "parallel_map" section.
 
-[Parallel DataFlow tutorial](parallel-dataflow.html) give a deeper dive
+[Parallel DataFlow tutorial](parallel-dataflow.html) gives a deeper dive
 on how to use them to optimize your data pipeline.
 
 ### Run the DataFlow

docs/tutorial/philosophy/dataflow.md

@@ -5,8 +5,6 @@ There are many other data loading solutions for deep learning.
 Here we explain why you may want to use Tensorpack DataFlow for your own good:
 **it's easy, and fast (enough)**.
 
-Note that this article may contain subjective opinions and we're happy to hear different voices.
-
 ### How Fast Do You Actually Need?
 
 Your data pipeline **only needs to be fast enough**.
@@ -42,7 +40,8 @@ And for us, we may optimize DataFlow even more, but we just haven't found the re
 
 Certain libraries advocate for a new binary data format (e.g., TFRecords, RecordIO).
 Do you need to use them?
-We think you usually do not, at least not after you try DataFlow, because they are:
+We think you usually do not, at least not after you try DataFlow, because these
+formats are:
 
 1. **Not Easy**: To use the new binary format,
 	 you need to write a script, to process your data from its original format,
@@ -98,7 +97,7 @@ Some frameworks have also provided good framework-specific solutions for data lo
 On the contrary, DataFlow is framework-agnostic: you can use it in any Python environment.
 In addition to this benefit, there are other reasons you might prefer DataFlow over the alternatives:
 
-#### tf.data or other TF operations
+#### tf.data and other graph operations
 
 The huge disadvantage of loading data in a computation graph is obvious:
 __it's extremely inflexible__.

examples/FasterRCNN/modeling/model_fpn.py

@@ -128,7 +128,7 @@ def multilevel_roi_align(features, rcnn_boxes, resolution):
     # Unshuffle to the original order, to match the original samples
     level_id_perm = tf.concat(level_ids, axis=0)  # A permutation of 1~N
     level_id_invert_perm = tf.invert_permutation(level_id_perm)
-    all_rois = tf.gather(all_rois, level_id_invert_perm)
+    all_rois = tf.gather(all_rois, level_id_invert_perm, name="output")
     return all_rois
 
 
@@ -202,7 +202,7 @@ def generate_fpn_proposals(
         # Detectron picks top-k within the batch, rather than within an image. However we do not have a batch.
         proposal_topk = tf.minimum(tf.size(proposal_scores), fpn_nms_topk)
         proposal_scores, topk_indices = tf.nn.top_k(proposal_scores, k=proposal_topk, sorted=False)
-        proposal_boxes = tf.gather(proposal_boxes, topk_indices)
+        proposal_boxes = tf.gather(proposal_boxes, topk_indices, name="all_proposals")
     else:
         for lvl in range(num_lvl):
             with tf.name_scope('Lvl{}'.format(lvl + 2)):

examples/FasterRCNN/train.py

@@ -31,8 +31,9 @@ if __name__ == '__main__':
     import multiprocessing as mp
     mp.set_start_method('spawn')
     parser = argparse.ArgumentParser()
-    parser.add_argument('--load', help='load a model to start training from. Can overwrite BACKBONE.WEIGHTS')
-    parser.add_argument('--logdir', help='log directory', default='train_log/maskrcnn')
+    parser.add_argument('--load', help='Load a model to start training from. It overwrites BACKBONE.WEIGHTS')
+    parser.add_argument('--logdir', help='Log directory. Will remove the old one if already exists.',
+                        default='train_log/maskrcnn')
     parser.add_argument('--config', help="A list of KEY=VALUE to overwrite those defined in config.py", nargs='+')
 
     if get_tf_version_tuple() < (1, 6):

tensorpack/callbacks/inference_runner.py

@@ -112,8 +112,8 @@ class InferenceRunner(InferenceRunnerBase):
             input (InputSource or DataFlow): The :class:`InputSource` to run
                 inference on.  If given a DataFlow, will use :class:`FeedInput`.
             infs (list): a list of :class:`Inferencer` instances.
-            tower_name (str): the name scope of the tower to build. Need to set a
-                different one if multiple InferenceRunner are used.
+            tower_name (str): the name scope of the tower to build.
+                If multiple InferenceRunner are used, each needs a different tower_name.
             tower_func (tfutils.TowerFunc or None): the tower function to be used to build the graph.
                 By defaults to call `trainer.tower_func` under a `training=False` TowerContext,
                 but you can change it to a different tower function
@@ -194,8 +194,8 @@ class DataParallelInferenceRunner(InferenceRunnerBase):
         Args:
             input (DataFlow or QueueInput)
             gpus (int or list[int]): #gpus, or list of GPU id
-            tower_name (str): the name scope of the tower to build. Need to set a
-                different one if multiple InferenceRunner are used.
+            tower_name (str): the name scope of the tower to build.
+                If multiple InferenceRunner are used, each needs a different tower_name.
             tower_func (tfutils.TowerFunc or None): the tower function to be used to build the graph.
                 The tower function will be called under a `training=False` TowerContext.
                 The default is `trainer.tower_func`,