Merge pull request rapidsai#169 from benfred/branch-24.08-merge-24.06

Forward-merge branch-24.06 into branch-24.08

.devcontainer/cuda11.8-conda/devcontainer.json

@@ -11,7 +11,7 @@
   "runArgs": [
     "--rm",
     "--name",
-    "${localEnv:USER}-rapids-${localWorkspaceFolderBasename}-24.08-cuda11.8-conda"
+    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-24.08-cuda11.8-conda"
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {

.devcontainer/cuda11.8-pip/devcontainer.json

@@ -11,7 +11,7 @@
   "runArgs": [
     "--rm",
     "--name",
-    "${localEnv:USER}-rapids-${localWorkspaceFolderBasename}-24.08-cuda11.8-pip"
+    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-24.08-cuda11.8-pip"
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {

.devcontainer/cuda12.2-conda/devcontainer.json

@@ -11,7 +11,7 @@
   "runArgs": [
     "--rm",
     "--name",
-    "${localEnv:USER}-rapids-${localWorkspaceFolderBasename}-24.08-cuda12.2-conda"
+    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-24.08-cuda12.2-conda"
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {

.devcontainer/cuda12.2-pip/devcontainer.json

@@ -11,7 +11,7 @@
   "runArgs": [
     "--rm",
     "--name",
-    "${localEnv:USER}-rapids-${localWorkspaceFolderBasename}-24.08-cuda12.2-pip"
+    "${localEnv:USER:anon}-rapids-${localWorkspaceFolderBasename}-24.08-cuda12.2-pip"
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,6 +1,6 @@
 <!--
 
-Thank you for contributing to RAFT :)
+Thank you for contributing to cuvs :)
 
 Here are some guidelines to help the review process go smoothly.
 

cpp/CMakeLists.txt

@@ -243,6 +243,7 @@ add_library(
   src/neighbors/cagra_serialize_float.cu
   src/neighbors/cagra_serialize_int8.cu
   src/neighbors/cagra_serialize_uint8.cu
+  src/neighbors/detail/cagra/cagra_build.cpp
   src/neighbors/detail/cagra/q_search_multi_cta_float_uint32_dim128_t8_8pq_2subd_half.cu
   src/neighbors/detail/cagra/q_search_multi_cta_float_uint32_dim128_t8_8pq_4subd_half.cu
   src/neighbors/detail/cagra/q_search_multi_cta_float_uint32_dim256_t16_8pq_2subd_half.cu
@@ -387,10 +388,6 @@ add_library(
   src/neighbors/detail/cagra/search_single_cta_half_uint64_dim256_t16.cu
   src/neighbors/detail/cagra/search_single_cta_half_uint64_dim512_t32.cu
   src/neighbors/detail/cagra/search_single_cta_half_uint64_dim1024_t32.cu
-  src/neighbors/detail/refine_host_float_float.cpp
-  src/neighbors/detail/refine_host_half_float.cpp
-  src/neighbors/detail/refine_host_int8_t_float.cpp
-  src/neighbors/detail/refine_host_uint8_t_float.cpp
   src/neighbors/ivf_flat_index.cpp
   src/neighbors/ivf_flat/ivf_flat_build_extend_float_int64_t.cu
   src/neighbors/ivf_flat/ivf_flat_build_extend_int8_t_int64_t.cu
@@ -429,10 +426,22 @@ add_library(
   src/neighbors/ivf_pq/detail/ivf_pq_search_with_filter_float_int64_t.cu
   src/neighbors/ivf_pq/detail/ivf_pq_search_with_filter_int8_t_int64_t.cu
   src/neighbors/ivf_pq/detail/ivf_pq_search_with_filter_uint8_t_int64_t.cu
+  src/neighbors/nn_descent.cu
   src/neighbors/nn_descent_float.cu
   src/neighbors/nn_descent_int8.cu
   src/neighbors/nn_descent_uint8.cu
+  src/neighbors/refine/detail/refine_device_float_float.cu
+  src/neighbors/refine/detail/refine_device_half_float.cu
+  src/neighbors/refine/detail/refine_device_int8_t_float.cu
+  src/neighbors/refine/detail/refine_device_uint8_t_float.cu
+  src/neighbors/refine/detail/refine_host_float_float.cpp
+  src/neighbors/refine/detail/refine_host_half_float.cpp
+  src/neighbors/refine/detail/refine_host_int8_t_float.cpp
+  src/neighbors/refine/detail/refine_host_uint8_t_float.cpp
   src/neighbors/sample_filter.cu
+  src/selection/select_k_float_int64_t.cu
+  src/selection/select_k_float_uint32_t.cu
+  src/selection/select_k_half_uint32_t.cu
 )
 
 target_compile_options(

cpp/doxygen/Doxyfile

@@ -454,7 +454,7 @@ LOOKUP_CACHE_SIZE      = 0
 # than 0 to get more control over the balance between CPU load and processing
 # speed. At this moment only the input processing can be done using multiple
 # threads. Since this is still an experimental feature the default is set to 1,
-# which efficively disables parallel processing. Please report any issues you
+# which effectively disables parallel processing. Please report any issues you
 # encounter. Generating dot graphs in parallel is controlled by the
 # DOT_NUM_THREADS setting.
 # Minimum value: 0, maximum value: 32, default value: 1.

cpp/include/cuvs/cluster/agglomerative.hpp

@@ -26,6 +26,11 @@ namespace cuvs::cluster::agglomerative {
 // constant to indirectly control the number of neighbors. k = sqrt(n) + c. default to 15
 constexpr int DEFAULT_CONST_C = 15;
 
+/**
+ * @defgroup agglomerative_params agglomerative clustering hyperparameters
+ * @{
+ */
+
 /**
  * Determines the method for computing the minimum spanning tree (MST)
  */
@@ -45,6 +50,10 @@ enum Linkage {
   KNN_GRAPH = 1
 };
 
+/**
+ * @}
+ */
+
 /**
  * Simple container object for consolidating linkage results. This closely
  * mirrors the trained instance variables populated in
@@ -75,6 +84,10 @@ class single_linkage_output {
   }
 };
 
+/**
+ * @defgroup single_linkage single-linkage clustering APIs
+ * @{
+ */
 /**
  * Single-linkage clustering, capable of constructing a KNN graph to
  * scale the algorithm beyond the n^2 memory consumption of implementations
@@ -105,4 +118,7 @@ void single_linkage(
   cuvs::cluster::agglomerative::Linkage linkage = cuvs::cluster::agglomerative::Linkage::KNN_GRAPH,
   std::optional<int> c                          = std::make_optional<int>(DEFAULT_CONST_C));
 
+/**
+ * @}
+ */
 };  // end namespace  cuvs::cluster::agglomerative