* 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 stateless_normal_proto.h
* \brief
*/
#ifndef OPS_BUILT_IN_OP_PROTO_INC_STATELESS_NORMAL_H_
#define OPS_BUILT_IN_OP_PROTO_INC_STATELESS_NORMAL_H_
#include "graph/operator.h"
#include "graph/operator_reg.h"
namespace ge {
* @brief Outputs deterministic pseudorandom values from a normal distribution,
* with GPU-parity (same seed+offset produces same sequence as CUDA). \n
* @par Inputs:
* @li shape: 1-D. The shape of the output tensor. Must be one of the following types: int64.
* @li seed: 0-D. Seed for the Philox4x32-10 RNG algorithm. Must be one of the following types: int64.
* @li offset: 0-D. Offset for the Philox4x32-10 RNG algorithm. Must be one of the following types: int64.
* @li mean: Scalar or tensor. Mean of the normal distribution. Must be one of the following types: float, float16, bfloat16.
* @li std: Scalar or tensor. Standard deviation of the normal distribution. Must be one of the following types: float, float16, bfloat16. \n
* @par Attributes:
* dtype: Output data type. Must be one of the following types: float16, bfloat16, float32.
* Defaults to float32. \n
* @par Outputs:
* y: Returns random values with specified shape.
* Must be one of the following types: float16, bfloat16, float32. \n
* @par Third-party framework compatibility
* Compatible with PyTorch torch.normal (stateless, GPU-parity mode).
*/
REG_OP(StatelessNormal)
.INPUT(shape, TensorType({DT_INT64}))
.INPUT(seed, TensorType({DT_INT64}))
.INPUT(offset, TensorType({DT_INT64}))
.INPUT(mean, TensorType({DT_FLOAT, DT_BF16, DT_FLOAT16}))
.INPUT(std, TensorType({DT_FLOAT, DT_BF16, DT_FLOAT16}))
.OUTPUT(y, TensorType({DT_FLOAT16, DT_BF16, DT_FLOAT}))
.ATTR(dtype, Type, DT_FLOAT)
.OP_END_FACTORY_REG(StatelessNormal)
}
#endif
