repo-sync-2024-08-26T11:04:20+0800 (#834)

# Pull Request

## What problem does this PR solve?

Issue Number: Fixed #

## Possible side effects?

- Performance:

- Backward compatibility:

libspu/dialect/pphlo/transforms/hlo_legalize_to_pphlo.cc

@@ -1263,10 +1263,9 @@ class HloToPPHloOpConverter<stablehlo::DynamicUpdateSliceOp>
     Type result_type = typetools_.getType(
         this->getTypeConverter()->convertType(op.getType()), result_vis);
 
-    auto materialized = materializeInputs(op, op->getOperands());
+    auto materialized = materializeInputs(op, adaptor.getOperands());
     rewriter.replaceOpWithNewOp<pphlo::DynamicUpdateSliceOp>(
-        op, result_type, materialized[0], materialized[1],
-        adaptor.getStartIndices());
+        op, TypeRange{result_type}, materialized);
 
     return success();
   }

sml/preprocessing/preprocessing.py

@@ -340,7 +340,7 @@ def fit(self, X, *, zero_variance=False, contain_nan=False):
 
         contain_nan : bool, default=False
             Set to True to handle the nan value.
-            This option desiced whether to use nanmin and nanmax to compute the minimum
+            This option decides whether to use nanmin and nanmax to compute the minimum
             and maximum.
         """
         self._reset()
@@ -364,7 +364,7 @@ def partial_fit(self, X, *, zero_variance=False, contain_nan=False):
 
         contain_nan : bool, default=False
             Set to True to handle the nan value.
-            This option desiced whether to use nanmin and nanmax to compute the minimum
+            This option decides whether to use nanmin and nanmax to compute the minimum
             and maximum.
         """
         feature_range = self.feature_range
@@ -434,7 +434,7 @@ def fit_transform(self, X, *, zero_variance=False, contain_nan=False):
 
         contain_nan : bool, default=False
             Set to True to handle the nan value.
-            This option desiced whether to use nanmin and nanmax to compute the minimum
+            This option decides whether to use nanmin and nanmax to compute the minimum
             and maximum.
 
         Returns
@@ -495,7 +495,7 @@ def fit(self, X, zero_maxabs=False, contain_nan=False):
 
         contain_nan : bool, default=False
             Set to True to handle the nan value.
-            This option desiced whether to use nanmin and nanmax to compute the minimum
+            This option decides whether to use nanmin and nanmax to compute the minimum
             and maximum.
         """
         self._reset()
@@ -519,7 +519,7 @@ def partial_fit(self, X, *, zero_maxabs=False, contain_nan=False):
 
         contain_nan : bool, default=False
             Set to True to handle the nan value.
-            This option desiced whether to use nanmin and nanmax to compute the minimum
+            This option decides whether to use nanmin and nanmax to compute the minimum
             and maximum.
         """
         first_pass = not hasattr(self, "n_samples_seen_")
@@ -569,7 +569,7 @@ def fit_transform(self, X, *, zero_maxabs=False, contain_nan=False):
 
         contain_nan : bool, default=False
             Set to True to handle the nan value.
-            This option desiced whether to use nanmin and nanmax to compute the minimum
+            This option decides whether to use nanmin and nanmax to compute the minimum
             and maximum.
 
         Returns
@@ -687,9 +687,9 @@ def loop_body(i, st):
 class KBinsDiscretizer:
     """Bin continuous data into intervals.
 
-    Attribute encode is not implemented, since there is currently no onehotencoder
+    Attribute encode is not implemented, since there is currently no OneHotEncoder
     in sml.
-    Attribute subsample is not implemented, since random choise in SPU runtime does
+    Attribute subsample is not implemented, since random choice in SPU runtime does
     not work as expected.
 
     Parameters
@@ -744,7 +744,7 @@ def fit(
         Note that there is currently no support for handling constant values in a feature
         , since it introduces much redundant boolean computation because dynamic shape is
         not supported.
-        In sklarn, feature with constant value will be replaced with 0 after transformation.
+        In sklearn, feature with constant value will be replaced with 0 after transformation.
         (see https://github.com/scikit-learn/scikit-learn/blob/d139ff234b0f8ec30287e26e0bc801bdafdfbb1a/sklearn/preprocessing/tests/test_discretization.py#L192)
 
         Parameters