Merge branch 'branch-0.16' of https://github.com/rapidsai/cuml into f…

…ea-ext-lobpcg
venkywonka · Sep 10, 2020 · e19f1b6 · e19f1b6
2 parents 9836fdd + d84283e
commit e19f1b6
Show file tree

Hide file tree

Showing 100 changed files with 595 additions and 443 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,9 +2,11 @@
 
 ## New Features
 - PR #2698: Distributed TF-IDF transformer
+- PR #2638: Improve cython build with custom `build_ext`
 - PR #2772: Linearly Optimal Block Preconditioned Conjugate Gradient (LOBPCG) Solver
 
 ## Improvements
+- PR #2796: Remove tokens of length 1 by default for text vectorizers
 - PR #2741: Use rapids build packages in conda environments
 - PR #2735: Update seed to random_state in random forest and associated tests
 - PR #2739: Use cusparse_wrappers.h from RAFT
@@ -13,10 +15,18 @@
 - PR #2762: Fix broken links and provide minor edits to docs
 - PR #2723: Support and enable convert_dtype in estimator predict
 - PR #2758: Match sklearn's default n_components behavior for PCA
+- PR #2770: Fix doxygen version during cmake
+- PR #2766: Update default RandomForestRegressor score function to use r2
+- PR #2783: Add pytest that will fail when GPU IDs in Dask cluster are not unique
+- PR #2785: Add in cuML-specific dev conda dependencies
+- PR #2778: Add README for FIL
+- PR #2799: Reenable lightgbm test with lower (1%) proba accuracy
+- PR #2800: Align cuML's spdlog version with RMM's
 
 ## Bug Fixes
 - PR #2744: Supporting larger number of classes in KNeighborsClassifier
 - PR #2769: Remove outdated doxygen options for 1.8.20
+- PR #2787: Skip lightgbm test for version 3 and above temporarily
 
 # cuML 0.15.0 (Date TBD)
 
@@ -111,6 +121,7 @@
 - PR #2623: Fixing kmeans score() API to be compatible with Scikit-learn
 - PR #2629: Add naive_bayes api docs
 - PR #2643: 'dense' and 'sparse' values of `storage_type` for FIL
+- PR #2691: Generic Base class attribute setter
 - PR #2666: Update MBSGD documentation to mention that the model is experimental
 - PR #2687: Update xgboost version to 1.2.0dev.rapidsai0.15
 - PR #2684: CUDA 11 conda development environment yml and faiss patch
@@ -366,6 +377,7 @@
 - PR #2305: Fixed race condition in DBScan
 - PR #2354: Fix broken links in README
 - PR #2619: Explicitly skip raft test folder for pytest 6.0.0
+- PR #2788: Set the minimum number of columns that can be sampled to 1 to fix 0 mem allocation error
 
 # cuML 0.13.0 (31 Mar 2020)
 

diff --git a/build.sh b/build.sh
@@ -226,10 +226,9 @@ fi
 if completeBuild || hasArg cuml || hasArg pydocs; then
     cd ${REPODIR}/python
     if [[ ${INSTALL_TARGET} != "" ]]; then
-        python setup.py build_ext -j${PARALLEL_LEVEL:-1} --inplace ${SINGLEGPU_PYTHON_FLAG}
-        python setup.py install --single-version-externally-managed --record=record.txt ${SINGLEGPU_PYTHON_FLAG}
+        python setup.py build_ext -j${PARALLEL_LEVEL:-1} ${SINGLEGPU_PYTHON_FLAG} install --single-version-externally-managed --record=record.txt
     else
-        python setup.py build_ext -j${PARALLEL_LEVEL:-1} --inplace --library-dir=${LIBCUML_BUILD_DIR} ${SINGLEGPU_PYTHON_FLAG}
+        python setup.py build_ext -j${PARALLEL_LEVEL:-1} --library-dir=${LIBCUML_BUILD_DIR} ${SINGLEGPU_PYTHON_FLAG}
     fi
 
     if hasArg pydocs; then

diff --git a/conda/environments/cuml_dev_cuda10.1.yml b/conda/environments/cuml_dev_cuda10.1.yml
@@ -15,6 +15,13 @@ dependencies:
 - dask-cudf=0.16.*
 - dask-cuda=0.16.*
 - ucx-py=0.16.*
+- dask-ml
+- doxygen>=1.8.20
+- libfaiss>=1.6.3
+- faiss-proc=*=cuda
+- umap-learn
+- scikit-learn=0.23.1
+- treelite=0.92
 - pip
 - pip:
     - sphinx_markdown_tables

diff --git a/conda/environments/cuml_dev_cuda10.2.yml b/conda/environments/cuml_dev_cuda10.2.yml
@@ -15,6 +15,13 @@ dependencies:
 - dask-cudf=0.16.*
 - dask-cuda=0.16.*
 - ucx-py=0.16.*
+- dask-ml
+- doxygen>=1.8.20
+- libfaiss>=1.6.3
+- faiss-proc=*=cuda
+- umap-learn
+- scikit-learn=0.23.1
+- treelite=0.92
 - pip
 - pip:
     - sphinx_markdown_tables

diff --git a/conda/environments/cuml_dev_cuda11.0.yml b/conda/environments/cuml_dev_cuda11.0.yml
@@ -15,6 +15,13 @@ dependencies:
 - dask-cudf=0.16.*
 - dask-cuda=0.16.*
 - ucx-py=0.16.*
+- dask-ml
+- doxygen>=1.8.20
+- libfaiss>=1.6.3
+- faiss-proc=*=cuda
+- umap-learn
+- scikit-learn=0.23.1
+- treelite=0.92
 - pip
 - pip:
     - sphinx_markdown_tables

diff --git a/cpp/cmake/Dependencies.cmake b/cpp/cmake/Dependencies.cmake
@@ -122,7 +122,7 @@ set(SPDLOG_DIR ${CMAKE_CURRENT_BINARY_DIR}/spdlog CACHE STRING
   "Path to spdlog install directory")
 ExternalProject_Add(spdlog
   GIT_REPOSITORY    https://github.com/gabime/spdlog.git
-  GIT_TAG           v1.x
+  GIT_TAG           v1.7.0
   PREFIX            ${SPDLOG_DIR}
   CONFIGURE_COMMAND ""
   BUILD_COMMAND     ""
@@ -254,4 +254,3 @@ add_dependencies(GTest::GTest spdlog)
 add_dependencies(benchmark GTest::GTest)
 add_dependencies(FAISS::FAISS benchmark)
 add_dependencies(FAISS::FAISS faiss)
-
diff --git a/cpp/cmake/doxygen.cmake b/cpp/cmake/doxygen.cmake
@@ -1,4 +1,4 @@
-# Copyright (c) 2019, NVIDIA CORPORATION.
+# Copyright (c) 2019-2020, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,7 +13,7 @@
 # limitations under the License.
 #
 
-find_package(Doxygen 1.8.11)
+find_package(Doxygen 1.8.20 REQUIRED)
 
 function(add_doxygen_target)
   if(Doxygen_FOUND)

diff --git a/cpp/src/decisiontree/memory.cuh b/cpp/src/decisiontree/memory.cuh
@@ -83,6 +83,7 @@ void TemporaryMemory<T, L>::LevelMemAllocator(
   }
   size_t maxnodes = max_nodes_per_level;
   size_t ncols_sampled = (size_t)(ncols * tree_params.max_features);
+  ncols_sampled = ncols_sampled > 0 ? ncols_sampled : 1;
   if (depth < 64) {
     gather_max_nodes = std::min((size_t)(nrows + 1),
                                 (size_t)(pow((size_t)2, (size_t)depth) + 1));

diff --git a/python/cuml/cluster/dbscan.pyx b/python/cuml/cluster/dbscan.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 import ctypes
 import cudf
@@ -217,8 +214,7 @@ class DBSCAN(Base):
             "int64", np.int64}.
 
         """
-        self._set_n_features_in(X)
-        self._set_output_type(X)
+        self._set_base_attributes(output_type=X, n_features=X)
 
         if self._labels_ is not None:
             del self._labels_

diff --git a/python/cuml/cluster/kmeans.pyx b/python/cuml/cluster/kmeans.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 import ctypes
 import cudf
@@ -313,8 +310,7 @@ class KMeans(Base):
         Compute k-means clustering with X.
 
         """
-        self._set_n_features_in(X)
-        self._set_output_type(X)
+        self._set_base_attributes(output_type=X, n_features=X)
 
         if self.init == 'preset':
             check_cols = self.n_cols

diff --git a/python/cuml/cluster/kmeans_mg.pyx b/python/cuml/cluster/kmeans_mg.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 import ctypes
 import cudf
@@ -87,7 +84,7 @@ class KMeansMG(KMeans):
             ndarray, cuda array interface compliant array like CuPy
 
         """
-        self._set_n_features_in(X)
+        self._set_base_attributes(n_features=X)
 
         X_m, self.n_rows, self.n_cols, self.dtype = \
             input_to_cuml_array(X, order='C')

diff --git a/python/cuml/cluster/kmeans_utils.pxd b/python/cuml/cluster/kmeans_utils.pxd
@@ -14,11 +14,6 @@
 # limitations under the License.
 #
 
-# cython: profile=False
-# distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
-
 import ctypes
 from libcpp cimport bool
 

diff --git a/python/cuml/common/base.pyx b/python/cuml/common/base.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 import cuml
 import cuml.common.cuda
@@ -282,12 +279,49 @@ class Base:
             else:
                 raise AttributeError
 
-    def _set_output_type(self, input):
+    def _set_base_attributes(self,
+                             output_type=None,
+                             target_dtype=None,
+                             n_features=None):
         """
-        Method to be called by fit methods of inheriting classes
-        to correctly set the output type depending on the type of inputs,
-        class output type and global output type
+        Method to set the base class attributes - output type,
+        target dtype and n_features. It combines the three different
+        function calls. It's called in fit function from estimators.
+
+        Parameters
+        --------
+        output_type : DataFrame (default = None)
+            Is output_type is passed, aets the output_type on the
+            dataframe passed
+        target_dtype : Target column (default = None)
+            If target_dtype is passed, we call _set_target_dtype
+            on it
+        n_features: int or DataFrame (default=None)
+            If an int is passed, we set it to the number passed
+            If dataframe, we set it based on the passed df.
+
+        Examples
+        --------
+
+        .. code-block:: python
+
+                # To set output_type and n_features based on X
+                self._set_base_attributes(output_type=X, n_features=X)
+
+                # To set output_type on X and n_features to 10
+                self._set_base_attributes(output_type=X, n_features=10)
+
+                # To only set target_dtype
+                self._set_base_attributes(output_type=X, target_dtype=y)
         """
+        if output_type is not None:
+            self._set_output_type(output_type)
+        if target_dtype is not None:
+            self._set_target_dtype(target_dtype)
+        if n_features is not None:
+            self._set_n_features_in(n_features)
+
+    def _set_output_type(self, input):
         if self.output_type == 'input' or self._mirror_input:
             self.output_type = _input_to_type(input)
 
@@ -303,11 +337,6 @@ class Base:
             return self.output_type
 
     def _set_target_dtype(self, target):
-        """
-        Method to be called by fit methods of inheriting classifier
-        classes to correctly set the output dtype depending on the dtype of
-        the target.
-        """
         self.target_dtype = _input_target_to_dtype(target)
 
     def _get_target_dtype(self):
@@ -323,9 +352,6 @@ class Base:
         return out_dtype
 
     def _set_n_features_in(self, X):
-        """Method to be called by the fit method of the inheriting class.
-        Sets the n_features_in_ attribute based on the data passed to fit.
-        """
         if isinstance(X, int):
             self.n_features_in_ = X
         else:

diff --git a/python/cuml/common/cuda.pxd b/python/cuml/common/cuda.pxd
@@ -14,12 +14,6 @@
 # limitations under the License.
 #
 
-# cython: profile=False
-# distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
-
-
 # Populate this with more typedef's (eg: events) as and when needed
 cdef extern from * nogil:
     ctypedef void* _Stream "cudaStream_t"

diff --git a/python/cuml/common/cuda.pyx b/python/cuml/common/cuda.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 import functools
 from libcpp.string cimport string

diff --git a/python/cuml/common/handle.pxd b/python/cuml/common/handle.pxd
@@ -14,12 +14,6 @@
 # limitations under the License.
 #
 
-# cython: profile=False
-# distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
-
-
 from libcpp.memory cimport shared_ptr
 cimport cuml.common.cuda
 

diff --git a/python/cuml/common/handle.pyx b/python/cuml/common/handle.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 
 import cuml

diff --git a/python/cuml/common/logger.pyx b/python/cuml/common/logger.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 
 from libcpp.string cimport string

diff --git a/python/cuml/common/opg_data_utils_mg.pxd b/python/cuml/common/opg_data_utils_mg.pxd
@@ -13,10 +13,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 #
-# cython: profile=False
-# distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 # Util functions, will be moved to their own file as the other methods are
 # refactored

diff --git a/python/cuml/common/pointer_utils.pyx b/python/cuml/common/pointer_utils.pyx
@@ -14,10 +14,7 @@
 # limitations under the License.
 #
 
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 from libc.stdint cimport uintptr_t
 

diff --git a/python/cuml/dask/common/comms_utils.pyx b/python/cuml/dask/common/comms_utils.pyx
@@ -12,10 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 #
-# cython: profile=False
 # distutils: language = c++
-# cython: embedsignature = True
-# cython: language_level = 3
 
 from libc.stdlib cimport malloc, free
 from cython.operator cimport dereference as deref