Update on "Add unsigned integer dtypes to PyTorch"

ezyang · ezyang · commit ced08cafa43a · 2024-01-03T20:35:19.000+08:00
The dtypes are very useless right now (not even fill works), but it makes torch.uint16, uint32 and uint64 available as a dtype. Towards #58734 Signed-off-by: Edward Z. Yang <ezyangmeta.com> [ghstack-poisoned]
diff --git a/c10/core/ScalarType.h b/c10/core/ScalarType.h
@@ -587,6 +587,8 @@ static inline bool canCast(const ScalarType from, const ScalarType to) {
   return true;
 }
 
+C10_API ScalarType promoteTypes(ScalarType a, ScalarType b);
+
 inline std::ostream& operator<<(
     std::ostream& stream,
     at::ScalarType scalar_type) {

Original file line number	Diff line number	Diff line change
`@@ -587,6 +587,8 @@ static inline bool canCast(const ScalarType from, const ScalarType to) {`
`587`	`587`	`return true;`
`588`	`588`	`}`
`589`	`589`
	`590`	`+C10_API ScalarType promoteTypes(ScalarType a, ScalarType b);`
	`591`	`+`
`590`	`592`	`inline std::ostream& operator<<(`
`591`	`593`	`std::ostream& stream,`
`592`	`594`	`at::ScalarType scalar_type) {`