[major] update docs

README.md

@@ -5,7 +5,13 @@ convolutions for point clouds on GPUs.
 
 ## Introduction
 
-As LiDAR sensors become increasingly popular, point clouds becomes more accessible
+Thanks to recent advances in 3D sensors, such as light detection and ranging 
+(LiDAR) sensors, 3D point clouds become increasingly accessible and widely 
+used in many important applications.
+Different from 2D dense images, 3D point clouds describe 3D objects that are 
+extremely sparse to their bounding space (usually less than 0.01% within the 
+bounding 3D volume).
+In this work, we 
 
 
 ## Installation

minuet/nn/functional/indexing.py

@@ -94,10 +94,10 @@ def query_sorted_index_with_offsets(
 
   Args:
     sources: the **sorted** input coordinates of shape
-      :math:`(N_\\text{in}, D)`
+      :math:`[N_\text{in}, D]`
     targets: the **sorted** output coordinates of shape
-      :math:`(N_\\text{out}, D)`
-    offsets: the offsets of the weights of shape :math:`(N_\\text{weight}, D)`
+      :math:`[N_\text{out}, D]`
+    offsets: the offsets of the weights of shape :math:`[N_\text{weight}, D]`
     source_batch_dims:
       the batch_dims tensor (of the input coordinates) if there are
       multiple coordinate tensors
@@ -106,8 +106,8 @@ def query_sorted_index_with_offsets(
       multiple coordinate tensors
 
   Returns:
-    A tensor specifying the kernel map of shape :math:`(N_\\text{weight}, N_\\text{out})`
-
+    A tensor specifying the kernel map of shape
+    :math:`[N_\text{weight}, N_\text{out}]`
   """
   if (source_batch_dims is None) != (target_batch_dims is None):
     raise ValueError("source_batch_dims and target_batch_dims must be both "

minuet/tensors.py

@@ -52,7 +52,7 @@ class SparseTensor(object):
   Note that a :py:class:`SparseTensor` can store multiple point clouds. This is
   achieved by the :py:attr:`batch_dims` tensor, where the indices of the points
   and the features of the :math:`i`-th point cloud are within the interval
-  :math:`[\\text{batch_dims}[i], \\text{batch_dims}[i + 1])`.
+  :math:`[\text{batch_dims}[i], \text{batch_dims}[i + 1])`.
 
   Args:
     features: the tensor that stores all features of the point clouds