Merge branch 'branch-24.04' into test-cuda-12.2

rapidsai · Feb 9, 2024 · da0c036 · da0c036
2 parents f94be82 + d2f509e
commit da0c036
Show file tree

Hide file tree

Showing 6 changed files with 13 additions and 10 deletions.
diff --git a/cpp/test/sg/fil_test.cu b/cpp/test/sg/fil_test.cu
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2023, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2024, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -787,6 +787,7 @@ class TreeliteFilTest : public BaseFilTest<real_t> {
         case fil::leaf_algo_t::GROVE_PER_CLASS_FEW_CLASSES:
         case fil::leaf_algo_t::GROVE_PER_CLASS_MANY_CLASSES: break;
       }
+      builder->EndNode();
     } else {
       int left          = root + 2 * (node - root) + 1;
       int right         = root + 2 * (node - root) + 2;
@@ -806,8 +807,6 @@ class TreeliteFilTest : public BaseFilTest<real_t> {
           }
         }
       }
-      node_to_treelite(builder, root, left);
-      node_to_treelite(builder, root, right);
       // TODO(levsnv): remove workaround once confirmed to work with empty category lists in
       // Treelite
       if (!right_categories.empty() && dense_node.is_categorical()) {
@@ -818,8 +817,10 @@ class TreeliteFilTest : public BaseFilTest<real_t> {
         adjust_threshold_to_treelite(&threshold, &left, &right, &default_left, this->ps.op);
         builder->NumericalTest(dense_node.fid(), threshold, default_left, this->ps.op, left, right);
       }
+      builder->EndNode();
+      node_to_treelite(builder, root, left);
+      node_to_treelite(builder, root, right);
     }
-    builder->EndNode();
   }
 
   void init_forest_impl(fil::forest_t<real_t>* pforest, fil::storage_type_t storage_type)
@@ -875,7 +876,7 @@ class TreeliteFilTest : public BaseFilTest<real_t> {
         postprocessor_name = "sigmoid";
       }
     } else if (this->ps.leaf_algo != fil::leaf_algo_t::FLOAT_UNARY_BINARY) {
-      postprocessor_name = "softmax";
+      postprocessor_name = "identity_multiclass";
       this->ps.output    = fil::output_t(this->ps.output | fil::output_t::SOFTMAX);
     } else if (this->ps.leaf_algo == GROVE_PER_CLASS) {
       postprocessor_name = "identity_multiclass";

diff --git a/python/cuml/benchmark/datagen.py b/python/cuml/benchmark/datagen.py
@@ -73,7 +73,7 @@ def _gen_data_regression(
     )
 
     X_df = cudf.DataFrame(X_arr)
-    y_df = cudf.Series(y_arr)
+    y_df = cudf.Series(np.squeeze(y_arr))
 
     return X_df, y_df
 

diff --git a/python/cuml/common/sparsefuncs.py b/python/cuml/common/sparsefuncs.py
@@ -160,8 +160,9 @@ def create_csr_matrix_from_count_df(
 
     doc_token_counts = count_df["doc_id"].value_counts().reset_index()
     del count_df
+
     doc_token_counts = doc_token_counts.rename(
-        {"doc_id": "token_counts", "index": "doc_id"}, axis=1
+        {"count": "token_counts"}, axis=1
     ).sort_values(by="doc_id")
 
     token_counts = _insert_zeros(

diff --git a/python/cuml/preprocessing/encoders.py b/python/cuml/preprocessing/encoders.py
@@ -498,7 +498,7 @@ def inverse_transform(self, X):
                 dropped_class_idx = Series(self.drop_idx_[feature])
                 dropped_class_mask = Series(cats).isin(cats[dropped_class_idx])
                 if len(cats) == 1:
-                    inv = Series(Index(cats[0]).repeat(X.shape[0]))
+                    inv = Series(Index([cats[0]]).repeat(X.shape[0]))
                     result[feature] = inv
                     continue
                 cats = cats[~dropped_class_mask]

diff --git a/python/cuml/tests/test_train_test_split.py b/python/cuml/tests/test_train_test_split.py
@@ -48,7 +48,7 @@ def test_split_dataframe(train_size, shuffle):
     assert all(X_test.index.to_pandas() == y_test.index.to_pandas())
 
     X_reconstructed = cudf.concat([X_train, X_test]).sort_values(by=["x"])
-    y_reconstructed = y_train.append(y_test).sort_values()
+    y_reconstructed = cudf.concat([y_train, y_test]).sort_values()
 
     assert all(X_reconstructed.reset_index(drop=True) == X)
     out = y_reconstructed.reset_index(drop=True).values_host == y.values_host
@@ -96,7 +96,7 @@ def test_split_column():
     )
 
     X_reconstructed = cudf.concat([X_train, X_test]).sort_values(by=["x"])
-    y_reconstructed = y_train.append(y_test).sort_values()
+    y_reconstructed = cudf.concat([y_train, y_test]).sort_values()
 
     assert all(
         data

diff --git a/python/pyproject.toml b/python/pyproject.toml
@@ -127,6 +127,7 @@ build-dir = "build/{wheel_tag}"
 cmake.build-type = "Release"
 cmake.minimum-version = "3.26.4"
 ninja.make-fallback = true
+sdist.exclude = ["*tests*"]
 sdist.reproducible = true
 wheel.packages = ["cuml"]