blocking int64 from reducing percision to int32

mindspore/ops/operations/random_ops.py

@@ -514,7 +514,8 @@ class UniformCandidateSampler(PrimitiveWithInfer):
             of num_sampled. If unique=True, num_sampled must be less than or equal to range_max.
         unique (bool): Whether all sampled classes in a batch are unique.
         range_max (int): The number of possible classes, must be non-negative.
-        seed (int): Random seed, must be non-negative. Default: 0.
+        seed (int): Used for random number generation, must be non-negative. If seed has a value of 0,
+            seed will be replaced with a randomly generated value. Default: 0.
         remove_accidental_hits (bool): Whether accidental hit is removed. Default: False.
 
     Inputs:
@@ -553,6 +554,7 @@ class UniformCandidateSampler(PrimitiveWithInfer):
         self.num_sampled = num_sampled
 
     def infer_dtype(self, true_classes_type):
+        Validator.check_tensor_dtype_valid("true_classes_type", true_classes_type, (mstype.int32), self.name)
         return (true_classes_type, mstype.float32, mstype.float32)
 
     def infer_shape(self, true_classes_shape):