wehub-resource-sync
41 lines
1.4 KiB
C++
41 lines
1.4 KiB
C++
// Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
|
|
//
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
#pragma once
|
|
|
|
#include "paddle/phi/common/scalar.h"
|
|
#include "paddle/phi/core/dense_tensor.h"
|
|
|
|
namespace phi {
|
|
|
|
template <typename T, typename Context>
|
|
void LarsMomentumKernel(
|
|
const Context& dev_ctx,
|
|
const std::vector<const DenseTensor*>& param,
|
|
const std::vector<const DenseTensor*>& velocity,
|
|
const std::vector<const DenseTensor*>& learning_rate,
|
|
const std::vector<const DenseTensor*>& grad,
|
|
const optional<std::vector<const DenseTensor*>>& master_param,
|
|
const std::vector<float>& weight_decay_arr,
|
|
float mu,
|
|
float lars_coeff,
|
|
float epsilon,
|
|
bool multi_precision,
|
|
float rescale_grad,
|
|
std::vector<DenseTensor*> param_out,
|
|
std::vector<DenseTensor*> velocity_out,
|
|
std::vector<DenseTensor*> master_param_out);
|
|
|
|
} // namespace phi
|