mindspore.nn ============= 神经网络Cell。 用于构建神经网络中的预定义构建块或计算单元。 动态shape的支持情况详见 `nn接口动态shape支持情况 `_ 。 MindSpore中 `mindspore.nn` 接口与上一版本相比，新增、删除和支持平台的变化信息请参考 `mindspore.nn API接口变更 `_ 。 基本构成单元 ------------ .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Cell mindspore.nn.GraphCell mindspore.nn.LossBase mindspore.nn.Optimizer 容器 ----------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.CellDict mindspore.nn.CellList mindspore.nn.SequentialCell 封装层 ----------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.DistributedGradReducer mindspore.nn.DynamicLossScaleUpdateCell mindspore.nn.FixedLossScaleUpdateCell mindspore.nn.ForwardValueAndGrad mindspore.nn.GetNextSingleOp mindspore.nn.MicroBatchInterleaved mindspore.nn.ParameterUpdate mindspore.nn.PipelineCell mindspore.nn.TimeDistributed mindspore.nn.TrainOneStepCell mindspore.nn.TrainOneStepWithLossScaleCell mindspore.nn.WithEvalCell mindspore.nn.WithLossCell 卷积神经网络层 -------------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Conv1d mindspore.nn.Conv1dTranspose mindspore.nn.Conv2d mindspore.nn.Conv2dTranspose mindspore.nn.Conv3d mindspore.nn.Conv3dTranspose mindspore.nn.Unfold 循环神经网络层 ----------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.RNN mindspore.nn.RNNCell mindspore.nn.GRU mindspore.nn.GRUCell mindspore.nn.LSTM mindspore.nn.LSTMCell Transformer层 ----------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.MultiheadAttention mindspore.nn.TransformerEncoderLayer mindspore.nn.TransformerDecoderLayer mindspore.nn.TransformerEncoder mindspore.nn.TransformerDecoder mindspore.nn.Transformer 嵌入层 ----------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Embedding mindspore.nn.EmbeddingLookup mindspore.nn.MultiFieldEmbeddingLookup 非线性激活函数层 ------------------ .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.CELU mindspore.nn.ELU mindspore.nn.FastGelu mindspore.nn.GELU mindspore.nn.GLU mindspore.nn.get_activation mindspore.nn.Hardtanh mindspore.nn.HShrink mindspore.nn.HSigmoid mindspore.nn.HSwish mindspore.nn.LeakyReLU mindspore.nn.LogSigmoid mindspore.nn.LogSoftmax mindspore.nn.LRN mindspore.nn.Mish mindspore.nn.Softsign mindspore.nn.PReLU mindspore.nn.ReLU mindspore.nn.ReLU6 mindspore.nn.RReLU mindspore.nn.SeLU mindspore.nn.SiLU mindspore.nn.Sigmoid mindspore.nn.Softmin mindspore.nn.Softmax mindspore.nn.Softmax2d mindspore.nn.SoftShrink mindspore.nn.Tanh mindspore.nn.Tanhshrink mindspore.nn.Threshold 线性层 ----------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Dense mindspore.nn.BiDense Dropout层 ----------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Dropout mindspore.nn.Dropout1d mindspore.nn.Dropout2d mindspore.nn.Dropout3d 归一化层 --------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.BatchNorm1d mindspore.nn.BatchNorm2d mindspore.nn.BatchNorm3d mindspore.nn.GroupNorm mindspore.nn.InstanceNorm1d mindspore.nn.InstanceNorm2d mindspore.nn.InstanceNorm3d mindspore.nn.LayerNorm mindspore.nn.SyncBatchNorm 池化层 -------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.AdaptiveAvgPool1d mindspore.nn.AdaptiveAvgPool2d mindspore.nn.AdaptiveAvgPool3d mindspore.nn.AdaptiveMaxPool1d mindspore.nn.AdaptiveMaxPool2d mindspore.nn.AdaptiveMaxPool3d mindspore.nn.AvgPool1d mindspore.nn.AvgPool2d mindspore.nn.AvgPool3d mindspore.nn.FractionalMaxPool3d mindspore.nn.LPPool1d mindspore.nn.LPPool2d mindspore.nn.MaxPool1d mindspore.nn.MaxPool2d mindspore.nn.MaxPool3d mindspore.nn.MaxUnpool1d mindspore.nn.MaxUnpool2d mindspore.nn.MaxUnpool3d 填充层 -------------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Pad mindspore.nn.ConstantPad1d mindspore.nn.ConstantPad2d mindspore.nn.ConstantPad3d mindspore.nn.ReflectionPad1d mindspore.nn.ReflectionPad2d mindspore.nn.ReflectionPad3d mindspore.nn.ReplicationPad1d mindspore.nn.ReplicationPad2d mindspore.nn.ReplicationPad3d mindspore.nn.ZeroPad2d 损失函数 -------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.BCELoss mindspore.nn.BCEWithLogitsLoss mindspore.nn.CosineEmbeddingLoss mindspore.nn.CrossEntropyLoss mindspore.nn.CTCLoss mindspore.nn.DiceLoss mindspore.nn.FocalLoss mindspore.nn.GaussianNLLLoss mindspore.nn.HingeEmbeddingLoss mindspore.nn.HuberLoss mindspore.nn.KLDivLoss mindspore.nn.L1Loss mindspore.nn.MarginRankingLoss mindspore.nn.MSELoss mindspore.nn.MultiClassDiceLoss mindspore.nn.MultilabelMarginLoss mindspore.nn.MultiLabelSoftMarginLoss mindspore.nn.MultiMarginLoss mindspore.nn.NLLLoss mindspore.nn.PoissonNLLLoss mindspore.nn.RMSELoss mindspore.nn.SampledSoftmaxLoss mindspore.nn.SmoothL1Loss mindspore.nn.SoftMarginLoss mindspore.nn.SoftmaxCrossEntropyWithLogits mindspore.nn.TripletMarginLoss 优化器 ------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.Adadelta mindspore.nn.Adagrad mindspore.nn.Adam mindspore.nn.AdaMax mindspore.nn.AdamOffload mindspore.nn.AdamWeightDecay mindspore.nn.AdaSumByDeltaWeightWrapCell mindspore.nn.AdaSumByGradWrapCell mindspore.nn.ASGD mindspore.nn.FTRL mindspore.nn.Lamb mindspore.nn.LARS mindspore.nn.LazyAdam mindspore.nn.Momentum mindspore.nn.ProximalAdagrad mindspore.nn.RMSProp mindspore.nn.Rprop mindspore.nn.SGD mindspore.nn.thor 动态学习率 ----------- LearningRateSchedule类 ^^^^^^^^^^^^^^^^^^^^^^^ 本模块中的动态学习率都是LearningRateSchedule的子类，将LearningRateSchedule的实例传递给优化器。在训练过程中，优化器以当前step为输入调用该实例，得到当前的学习率。 .. code-block:: import mindspore.nn as nn min_lr = 0.01 max_lr = 0.1 decay_steps = 4 cosine_decay_lr = nn.CosineDecayLR(min_lr, max_lr, decay_steps) net = Net() optim = nn.Momentum(net.trainable_params(), learning_rate=cosine_decay_lr, momentum=0.9) .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.CosineDecayLR mindspore.nn.ExponentialDecayLR mindspore.nn.InverseDecayLR mindspore.nn.NaturalExpDecayLR mindspore.nn.PolynomialDecayLR mindspore.nn.WarmUpLR Dynamic LR函数 ^^^^^^^^^^^^^^ 本模块中的动态学习率都是function，调用function并将结果传递给优化器。在训练过程中，优化器将result[current step]作为当前学习率。 .. code-block:: import mindspore.nn as nn min_lr = 0.01 max_lr = 0.1 total_step = 6 step_per_epoch = 1 decay_epoch = 4 lr= nn.cosine_decay_lr(min_lr, max_lr, total_step, step_per_epoch, decay_epoch) net = Net() optim = nn.Momentum(net.trainable_params(), learning_rate=lr, momentum=0.9) .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.cosine_decay_lr mindspore.nn.exponential_decay_lr mindspore.nn.inverse_decay_lr mindspore.nn.natural_exp_decay_lr mindspore.nn.piecewise_constant_lr mindspore.nn.polynomial_decay_lr mindspore.nn.warmup_lr 图像处理层 ----------- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.PixelShuffle mindspore.nn.PixelUnshuffle mindspore.nn.ResizeBilinear mindspore.nn.Upsample 工具 ----- .. mscnplatformautosummary:: :toctree: nn :nosignatures: :template: classtemplate.rst mindspore.nn.ChannelShuffle mindspore.nn.Flatten mindspore.nn.Identity mindspore.nn.Unflatten