https://github.com/Microsoft/CNTK
Raw File
Tip revision: 2cd35ae89acd8888796e6169093cbd074d96f4cf authored by Amit Agarwal (ATISG) on 19 January 2017, 23:37:08 UTC
CNTK v2 library: Add the ability to broadcast an input without MBLayout along a specified layout, in the ReconcileDynamicAxis node
Tip revision: 2cd35ae
DistributedLearnerBase.h
//
// Copyright (c) Microsoft. All rights reserved.
// Licensed under the MIT license. See LICENSE.md file in the project root for full license information.
//

#pragma once

#include "CNTKLibrary.h"

namespace CNTK
{
    ///
    /// Base class for distributed learners.
    ///
    class DistributedLearnerBase : public DistributedLearner
    {
    public:
        Dictionary CreateCheckpoint() override;

        void RestoreFromCheckpoint(const Dictionary& checkpoint) override;

    protected:
        DistributedLearnerBase(DistributedCommunicatorPtr communicator, LearnerPtr learner, size_t distributeAfterSamples);

        static void PrepaireZeroGradients(std::unordered_map<Parameter, NDArrayViewPtr>& gradientValues, MinibatchInfo& info);
        static void ConvertToOrdered(const std::unordered_map<Parameter, NDArrayViewPtr>& gradientValues, std::vector<std::pair<Parameter, NDArrayViewPtr>>& result);

        std::vector<std::pair<Parameter, NDArrayViewPtr>> m_gradientBuffer;
        std::vector<Parameter> m_parameters;

        DistributedLearnerBase(const DistributedLearnerBase&) = delete; DistributedLearnerBase& operator=(const DistributedLearnerBase&) = delete; DistributedLearnerBase& operator=(DistributedLearnerBase&&) = delete; DistributedLearnerBase(DistributedLearnerBase&& other) = delete;
    };
}
back to top