@@ -167,7 +167,7 @@ def SparseTensorEncodingAttr : SparseTensor_Attr<"SparseTensorEncoding",
167
167
- **soa** : only applicable to singleton levels, fuses the singleton
168
168
level in SoA (structure of arrays) scheme.
169
169
170
- In addition to the map, the following two fields are optional:
170
+ In addition to the map, the following four fields are optional:
171
171
172
172
- The required bitwidth for position storage (integral offsets
173
173
into the sparse storage scheme). A narrow width reduces the memory
@@ -183,11 +183,11 @@ def SparseTensorEncodingAttr : SparseTensor_Attr<"SparseTensorEncoding",
183
183
coordinate over all levels). The choices are `8`, `16`, `32`,
184
184
`64`, or, the default, `0` to indicate a native bitwidth.
185
185
186
- - The required explicit value for the sparse tensor. If explicitVal is set,
186
+ - The explicit value for the sparse tensor. If explicitVal is set,
187
187
then all the non-zero values in the tensor have the same explicit value.
188
188
The default value Attribute() indicates that it is not set.
189
189
190
- - The required implicit value for the sparse tensor. If implicitVal is set,
190
+ - The implicit value for the sparse tensor. If implicitVal is set,
191
191
then the "zero" value in the tensor is equal to the implicit value.
192
192
For now, we only support `0` as the implicit value but it could be
193
193
extended in the future. The default value Attribute() indicates that
0 commit comments