* Copyright (c) 2025 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
* \file pow_proto.h
* \brief
*/
#ifndef OP_PROTO_POW_PROTO_H_
#define OP_PROTO_POW_PROTO_H_
#include "graph/operator_reg.h"
#include "graph/operator.h"
namespace ge {
* @brief Computes the pow of a tensor.
*@par Inputs:
* @li x: A tensor of type float16, float32, bfloat16, int8, uint8, int16, int32.
* @li exponent: A tensor of type float16, float32, bfloat16, int8, uint8, int16, int32.
*@par Outputs:
* y: A tensor of type float16, float32, bfloat16, int8, uint8, int16, int32.
*@par Third-party framework compatibility
* Compatible with the Pytorch operator Pow.
*@par Restrictions:
* @li x and exponent must have the same dtype.
* @li x and exponent must be broadcast-compatible (following PyTorch's broadcasting rules).
* @li The output tensor y must have the same shape as the broadcast result of x and exponent.
* @li If x is integer type and exponent is negative, the behavior is undefined (consistent with PyTorch).
*/
#define POW_TYPES \
DT_BF16, DT_FLOAT, DT_FLOAT16, DT_INT8, DT_UINT8, DT_INT16, DT_INT32
REG_OP(Pow)
.INPUT(x, TensorType({POW_TYPES}))
.INPUT(exponent, TensorType({POW_TYPES}))
.OUTPUT(y, TensorType({POW_TYPES}))
.OP_END_FACTORY_REG(Pow);
}
#endif
