package caffe

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: LayerParameter

• optional bool out_max_val = 1

  If true produce pairs (argmax, maxval)

• optional uint32 top_k = 2

• optional int32 axis = 3

  The axis along which to maximise -- may be negative to index from the end (e.g., -1 for the last axis). By default ArgMaxLayer maximizes over the flattened trailing dimensions for each index of the first / num dimension.

Used in: LayerParameter

• optional FillerParameter slope_filler = 1

• optional FillerParameter bias_filler = 2

• optional float momentum = 3

• optional float eps = 4

• optional bool frozen = 5

  If true, will use the moving average mean and std for training and test. Will override the lr_param and freeze all the parameters. Make sure to initialize the layer properly with pretrained parameters.

• optional BNParameter.Engine engine = 6

Used in: BNParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: LayerParameter

• optional bool use_global_stats = 1

  If false, accumulate global mean/variance values via a moving average. If true, use those accumulated values instead of computing mean/variance across the batch.

• optional float moving_average_fraction = 2

  How much does the moving average decay each iteration?

• optional float eps = 3

  Small value to add to the variance estimate so that we don't divide by zero.

Used in: LayerParameter

• optional int32 axis = 1

  The first axis of bottom[0] (the first input Blob) along which to apply bottom[1] (the second input Blob). May be negative to index from the end (e.g., -1 for the last axis). For example, if bottom[0] is 4D with shape 100x3x40x60, the output top[0] will have the same shape, and bottom[1] may have any of the following shapes (for the given value of axis): (axis == 0 == -4) 100; 100x3; 100x3x40; 100x3x40x60 (axis == 1 == -3) 3; 3x40; 3x40x60 (axis == 2 == -2) 40; 40x60 (axis == 3 == -1) 60 Furthermore, bottom[1] may have the empty shape (regardless of the value of "axis") -- a scalar bias.

• optional int32 num_axes = 2

  (num_axes is ignored unless just one bottom is given and the bias is a learned parameter of the layer. Otherwise, num_axes is determined by the number of axes by the second bottom.) The number of axes of the input (bottom[0]) covered by the bias parameter, or -1 to cover all axes of bottom[0] starting from `axis`. Set num_axes := 0, to add a zero-axis Blob: a scalar.

• optional FillerParameter filler = 3

  (filler is ignored unless just one bottom is given and the bias is a learned parameter of the layer.) The initialization for the learned bias parameter. Default is the zero (0) initialization, resulting in the BiasLayer initially performing the identity operation.

Used in: BlobProtoVector, LayerParameter, V1LayerParameter

• optional BlobShape shape = 7

• repeated float data = 5

• repeated float diff = 6

• repeated double double_data = 8

• repeated double double_diff = 9

• optional int32 num = 1

  4D dimensions -- deprecated. Use "shape" instead.

• optional int32 channels = 2

• optional int32 height = 3

• optional int32 width = 4

The BlobProtoVector is simply a way to pass multiple blobproto instances around.

• repeated BlobProto blobs = 1

Specifies the shape (dimensions) of a Blob.

Used in: BlobProto, InputParameter, NetParameter, ParameterParameter, ReshapeParameter

• repeated int64 dim = 1

Used in: LayerParameter, V1LayerParameter

• optional int32 axis = 2

  The axis along which to concatenate -- may be negative to index from the end (e.g., -1 for the last axis). Other axes must have the same dimension for all the bottom blobs. By default, ConcatLayer concatenates blobs along the "channels" axis (1).

• optional uint32 concat_dim = 1

  DEPRECATED: alias for "axis" -- does not support negative indexing.

Used in: LayerParameter, V1LayerParameter

• optional uint32 num_output = 1

  The number of outputs for the layer

• optional bool bias_term = 2

  whether to have bias terms

• repeated uint32 pad = 3

  Pad, kernel size, and stride are all given as a single value for equal dimensions in all spatial dimensions, or once per spatial dimension.

  The padding size; defaults to 0

• repeated uint32 kernel_size = 4

  The kernel size

• repeated uint32 stride = 6

  The stride; defaults to 1

• repeated uint32 dilation = 18

  Factor used to dilate the kernel, (implicitly) zero-filling the resulting holes. (Kernel dilation is sometimes referred to by its use in the algorithme à trous from Holschneider et al. 1987.)

  The dilation; defaults to 1

• optional uint32 pad_h = 9

  For 2D convolution only, the *_h and *_w versions may also be used to specify both spatial dimensions.

  The padding height (2D only)

• optional uint32 pad_w = 10

  The padding width (2D only)

• optional uint32 kernel_h = 11

  The kernel height (2D only)

• optional uint32 kernel_w = 12

  The kernel width (2D only)

• optional uint32 stride_h = 13

  The stride height (2D only)

• optional uint32 stride_w = 14

  The stride width (2D only)

• optional uint32 group = 5

  The group size for group conv

• optional FillerParameter weight_filler = 7

  The filler for the weight

• optional FillerParameter bias_filler = 8

  The filler for the bias

• optional ConvolutionParameter.Engine engine = 15

• optional int32 axis = 16

  The axis to interpret as "channels" when performing convolution. Preceding dimensions are treated as independent inputs; succeeding dimensions are treated as "spatial". With (N, C, H, W) inputs, and axis == 1 (the default), we perform N independent 2D convolutions, sliding C-channel (or (C/g)-channels, for groups g>1) filters across the spatial axes (H, W) of the input. With (N, C, D, H, W) inputs, and axis == 1, we perform N independent 3D convolutions, sliding (C/g)-channels filters across the spatial axes (D, H, W) of the input.

• optional bool force_nd_im2col = 17

  Whether to force use of the general ND convolution, even if a specific implementation for blobs of the appropriate number of spatial dimensions is available. (Currently, there is only a 2D-specific convolution implementation; for input blobs with num_axes != 2, this option is ignored and the ND implementation will be used.)

Used in: ConvolutionParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: LayerParameter

• optional int32 axis = 1

  To crop, elements of the first bottom are selected to fit the dimensions of the second, reference bottom. The crop is configured by - the crop `axis` to pick the dimensions for cropping - the crop `offset` to set the shift for all/each dimension to align the cropped bottom with the reference bottom. All dimensions up to but excluding `axis` are preserved, while the dimensions including and trailing `axis` are cropped. If only one `offset` is set, then all dimensions are offset by this amount. Otherwise, the number of offsets must equal the number of cropped axes to shift the crop in each dimension accordingly. Note: standard dimensions are N,C,H,W so the default is a spatial crop, and `axis` may be negative to index from the end (e.g., -1 for the last axis).

• repeated uint32 offset = 2

Message that store parameters used by DetectionOutputLayer

Used in: LayerParameter

• optional uint32 num_classes = 1

  Number of classes to be predicted. Required!

• optional bool share_location = 2

  If true, bounding box are shared among different classes.

• optional int32 background_label_id = 3

  Background label id. If there is no background class, set it as -1.

• optional NonMaximumSuppressionParameter nms_param = 4

  Parameters used for non maximum suppression.

• optional PriorBoxParameter.CodeType code_type = 6

  Type of coding method for bbox.

• optional bool variance_encoded_in_target = 8

  If true, variance is encoded in target; otherwise we need to adjust the predicted offset accordingly.

• optional int32 keep_top_k = 7

  Number of total bboxes to be kept per image after nms step. -1 means keeping all bboxes after nms step.

• optional float confidence_threshold = 9

  Only consider detections whose confidences are larger than a threshold. If not provided, consider all boxes.

• optional bool visualize = 10

  If true, visualize the detection results.

• optional float visualize_threshold = 11

  The threshold used to visualize the detection results.

• optional string save_file = 12

  If provided, save outputs to video file.

Used in: LayerParameter, V1LayerParameter

• optional float dropout_ratio = 1

  dropout ratio

Message that stores parameters used by ELULayer

Used in: LayerParameter

• optional float alpha = 1

  Described in: Clevert, D.-A., Unterthiner, T., & Hochreiter, S. (2015). Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs). arXiv

Used in: LayerParameter, V1LayerParameter

• optional EltwiseParameter.EltwiseOp operation = 1

  element-wise operation

• repeated float coeff = 2

  blob-wise coefficient for SUM operation

• optional bool stable_prod_grad = 3

  Whether to use an asymptotically slower (for >2 inputs) but stabler method of computing the gradient for the PROD operation. (No effect for SUM op.)

Used in: EltwiseParameter

• PROD = 0

• SUM = 1

• MAX = 2

Message that stores parameters used by EmbedLayer

Used in: LayerParameter

• optional uint32 num_output = 1

  The number of outputs for the layer

• optional uint32 input_dim = 2

  The input is given as integers to be interpreted as one-hot vector indices with dimension num_input. Hence num_input should be 1 greater than the maximum possible input value.

• optional bool bias_term = 3

  Whether to use a bias term

• optional FillerParameter weight_filler = 4

  The filler for the weight

• optional FillerParameter bias_filler = 5

  The filler for the bias

Message that stores parameters used by ExpLayer

Used in: LayerParameter, V1LayerParameter

• optional float base = 1

  ExpLayer computes outputs y = base ^ (shift + scale * x), for base > 0. Or if base is set to the default (-1), base is set to e, so y = exp(shift + scale * x).

• optional float scale = 2

• optional float shift = 3

Used in: BNParameter, BiasParameter, ConvolutionParameter, EmbedParameter, InnerProductParameter, NormalizeParameter, PReLUParameter, ScaleParameter

• optional string type = 1

  The filler type.

• optional float value = 2

  the value in constant filler

• optional float min = 3

  the min value in uniform filler

• optional float max = 4

  the max value in uniform filler

• optional float mean = 5

  the mean value in Gaussian filler

• optional float std = 6

  the std value in Gaussian filler

• optional int32 sparse = 7

  The expected number of non-zero output weights for a given input in Gaussian filler -- the default -1 means don't perform sparsification.

• optional FillerParameter.VarianceNorm variance_norm = 8

Normalize the filler variance by fan_in, fan_out, or their average. Applies to 'xavier' and 'msra' fillers.

Used in: FillerParameter

• FAN_IN = 0

• FAN_OUT = 1

• AVERAGE = 2

/ Message that stores parameters used by FlattenLayer

Used in: LayerParameter

• optional int32 axis = 1

  The first axis to flatten: all preceding axes are retained in the output. May be negative to index from the end (e.g., -1 for the last axis).

• optional int32 end_axis = 2

  The last axis to flatten: all following axes are retained in the output. May be negative to index from the end (e.g., the default -1 for the last axis).

Used in: LayerParameter, V1LayerParameter

• optional uint32 num_output = 1

  The number of outputs for the layer

• optional bool bias_term = 2

  whether to have bias terms

• optional FillerParameter weight_filler = 3

  The filler for the weight

• optional FillerParameter bias_filler = 4

  The filler for the bias

• optional int32 axis = 5

  The first axis to be lumped into a single inner product computation; all preceding axes are retained in the output. May be negative to index from the end (e.g., -1 for the last axis).

• optional bool transpose = 6

  Specify whether to transpose the weight matrix or not. If transpose == true, any operations will be performed on the transpose of the weight matrix. The weight matrix itself is not going to be transposed but rather the transfer flag of operations will be toggled accordingly.

Used in: LayerParameter

• repeated BlobShape shape = 1

  This layer produces N >= 1 top blob(s) to be assigned manually. Define N shapes to set a shape for each top. Define 1 shape to set the same shape for every top. Define no shape to defer to reshaping manually.

Message that stores parameters used by LRNLayer

Used in: LayerParameter, V1LayerParameter

• optional uint32 local_size = 1

• optional float alpha = 2

• optional float beta = 3

• optional LRNParameter.NormRegion norm_region = 4

• optional float k = 5

• optional LRNParameter.Engine engine = 6

Used in: LRNParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: LRNParameter

• ACROSS_CHANNELS = 0

• WITHIN_CHANNEL = 1

NOTE Update the next available ID when you add a new LayerParameter field. LayerParameter next available layer-specific ID: 147 (last added: recurrent_param)

Used in: NetParameter

• optional string name = 1

  the layer name

• optional string type = 2

  the layer type

• repeated string bottom = 3

  the name of each bottom blob

• repeated string top = 4

  the name of each top blob

• optional Phase phase = 10

  The train / test phase for computation.

• repeated float loss_weight = 5

  The amount of weight to assign each top blob in the objective. Each layer assigns a default value, usually of either 0 or 1, to each top blob.

• repeated ParamSpec param = 6

  Specifies training parameters (multipliers on global learning constants, and the name and other settings used for weight sharing).

• repeated BlobProto blobs = 7

  The blobs containing the numeric parameters of the layer.

• repeated bool propagate_down = 11

  Specifies whether to backpropagate to each bottom. If unspecified, Caffe will automatically infer whether each input needs backpropagation to compute parameter gradients. If set to true for some inputs, backpropagation to those inputs is forced; if set false for some inputs, backpropagation to those inputs is skipped. The size must be either 0 or equal to the number of bottoms.

• repeated NetStateRule include = 8

  Rules controlling whether and when a layer is included in the network, based on the current NetState. You may specify a non-zero number of rules to include OR exclude, but not both. If no include or exclude rules are specified, the layer is always included. If the current NetState meets ANY (i.e., one or more) of the specified rules, the layer is included/excluded.

• repeated NetStateRule exclude = 9

• optional LossParameter loss_param = 200

  Parameters shared by loss layers.

• optional ArgMaxParameter argmax_param = 201

  Layer type-specific parameters. Note: certain layers may have more than one computational engine for their implementation. These layers include an Engine type and engine parameter for selecting the implementation. The default for the engine is set by the ENGINE switch at compile-time.

• optional BatchNormParameter batch_norm_param = 202

• optional BiasParameter bias_param = 203

• optional BNParameter bn_param = 204

• optional ConcatParameter concat_param = 205

• optional ConvolutionParameter convolution_param = 206

• optional CropParameter crop_param = 207

• optional DropoutParameter dropout_param = 208

• optional EltwiseParameter eltwise_param = 209

• optional ELUParameter elu_param = 210

• optional EmbedParameter embed_param = 211

• optional ExpParameter exp_param = 212

• optional FlattenParameter flatten_param = 213

• optional InnerProductParameter inner_product_param = 214

• optional InputParameter input_param = 215

• optional LogParameter log_param = 216

• optional LRNParameter lrn_param = 217

• optional MVNParameter mvn_param = 218

• optional ParameterParameter parameter_param = 219

• optional PoolingParameter pooling_param = 220

• optional PowerParameter power_param = 222

• optional PReLUParameter prelu_param = 223

• optional ProposalParameter proposal_param = 224

• optional PSROIPoolingParameter psroi_pooling_param = 225

• optional ReductionParameter reduction_param = 226

• optional ReLUParameter relu_param = 227

• optional ReshapeParameter reshape_param = 228

• optional ROIPoolingParameter roi_pooling_param = 229

• optional ScaleParameter scale_param = 230

• optional SigmoidParameter sigmoid_param = 231

• optional SoftmaxParameter softmax_param = 232

• optional SPPParameter spp_param = 233

• optional SliceParameter slice_param = 234

• optional TanHParameter tanh_param = 235

• optional ThresholdParameter threshold_param = 236

• optional TileParameter tile_param = 237

• optional DetectionOutputParameter detection_output_param = 238

• optional PermuteParameter permute_param = 239

• optional PriorBoxParameter prior_box_param = 240

• optional ShuffleChannelParameter shuffle_channel_param = 241

• optional NormalizeParameter norm_param = 242

Message that stores parameters used by LogLayer

Used in: LayerParameter

• optional float base = 1

  LogLayer computes outputs y = log_base(shift + scale * x), for base > 0. Or if base is set to the default (-1), base is set to e, so y = ln(shift + scale * x) = log_e(shift + scale * x)

• optional float scale = 2

• optional float shift = 3

Message that stores parameters shared by loss layers

Used in: LayerParameter, V1LayerParameter

• optional int32 ignore_label = 1

  If specified, ignore instances with the given label.

• optional LossParameter.NormalizationMode normalization = 3

• optional bool normalize = 2

  Deprecated. Ignored if normalization is specified. If normalization is not specified, then setting this to false will be equivalent to normalization = BATCH_SIZE to be consistent with previous behavior.

How to normalize the loss for loss layers that aggregate across batches, spatial dimensions, or other dimensions. Currently only implemented in SoftmaxWithLoss layer.

Used in: LossParameter

• FULL = 0

  Divide by the number of examples in the batch times spatial dimensions. Outputs that receive the ignore label will NOT be ignored in computing the normalization factor.

• VALID = 1

  Divide by the total number of output locations that do not take the ignore_label. If ignore_label is not set, this behaves like FULL.

• BATCH_SIZE = 2

  Divide by the batch size.

• NONE = 3

  Do not normalize the loss.

Used in: LayerParameter, V1LayerParameter

• optional bool normalize_variance = 1

  This parameter can be set to false to normalize mean only

• optional bool across_channels = 2

  This parameter can be set to true to perform DNN-like MVN

• optional float eps = 3

  Epsilon for not dividing by zero while normalizing variance

• optional string name = 1

  consider giving the network a name

• repeated string input = 3

  DEPRECATED. See InputParameter. The input blobs to the network.

• repeated BlobShape input_shape = 8

  DEPRECATED. See InputParameter. The shape of the input blobs.

• repeated int32 input_dim = 4

  4D input dimensions -- deprecated. Use "input_shape" instead. If specified, for each input blob there should be four values specifying the num, channels, height and width of the input blob. Thus, there should be a total of (4 * #input) numbers.

• optional bool force_backward = 5

  Whether the network will force every layer to carry out backward operation. If set False, then whether to carry out backward is determined automatically according to the net structure and learning rates.

• optional NetState state = 6

  The current "state" of the network, including the phase, level, and stage. Some layers may be included/excluded depending on this state and the states specified in the layers' include and exclude fields.

• optional bool debug_info = 7

  Print debugging information about results while running Net::Forward, Net::Backward, and Net::Update.

• repeated LayerParameter layer = 100

  The layers that make up the net. Each of their configurations, including connectivity and behavior, is specified as a LayerParameter.

  ID 100 so layers are printed last.

• repeated V1LayerParameter layers = 2

  DEPRECATED: use 'layer' instead.

Used in: NetParameter

• optional Phase phase = 1

• optional int32 level = 2

• repeated string stage = 3

Used in: LayerParameter, V1LayerParameter

• optional Phase phase = 1

  Set phase to require the NetState have a particular phase (TRAIN or TEST) to meet this rule.

• optional int32 min_level = 2

  Set the minimum and/or maximum levels in which the layer should be used. Leave undefined to meet the rule regardless of level.

• optional int32 max_level = 3

• repeated string stage = 4

  Customizable sets of stages to include or exclude. The net must have ALL of the specified stages and NONE of the specified "not_stage"s to meet the rule. (Use multiple NetStateRules to specify conjunctions of stages.)

• repeated string not_stage = 5

Used in: DetectionOutputParameter

• optional float nms_threshold = 1

  Threshold to be used in nms.

• optional int32 top_k = 2

  Maximum number of results to be kept.

• optional float eta = 3

  Parameter for adaptive nms.

Message that stores parameters used by NormalizeLayer

Used in: LayerParameter

• optional bool across_spatial = 1

• optional FillerParameter scale_filler = 2

  Initial value of scale. Default is 1.0 for all

• optional bool channel_shared = 3

  Whether or not scale parameters are shared across channels.

• optional float eps = 4

  Epsilon for not dividing by zero while normalizing variance

Parametric ReLU described in K. He et al, Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification, 2015.

Used in: LayerParameter

• optional FillerParameter filler = 1

  Initial value of a_i. Default is a_i=0.25 for all i.

• optional bool channel_shared = 2

  Whether or not slope paramters are shared across channels.

Used in: LayerParameter

• required float spatial_scale = 1

• required int32 output_dim = 2

  output channel number

• required int32 group_size = 3

  equal to pooled_size

Specifies training parameters (multipliers on global learning constants, and the name and other settings used for weight sharing).

Used in: LayerParameter

• optional string name = 1

  The names of the parameter blobs -- useful for sharing parameters among layers, but never required otherwise. To share a parameter between two layers, give it a (non-empty) name.

• optional ParamSpec.DimCheckMode share_mode = 2

  Whether to require shared weights to have the same shape, or just the same count -- defaults to STRICT if unspecified.

• optional float lr_mult = 3

  The multiplier on the global learning rate for this parameter.

• optional float decay_mult = 4

  The multiplier on the global weight decay for this parameter.

Used in: ParamSpec

• STRICT = 0

  STRICT (default) requires that num, channels, height, width each match.

• PERMISSIVE = 1

  PERMISSIVE requires only the count (num*channels*height*width) to match.

Used in: LayerParameter

• optional BlobShape shape = 1

Used in: LayerParameter

• repeated uint32 order = 1

  The new orders of the axes of data. Notice it should be with in the same range as the input data, and it starts from 0. Do not provide repeated order.

Used in: LayerParameter, NetState, NetStateRule

• TRAIN = 0

• TEST = 1

Used in: LayerParameter, V1LayerParameter

• optional PoolingParameter.PoolMethod pool = 1

  The pooling method

• optional uint32 pad = 4

  Pad, kernel size, and stride are all given as a single value for equal dimensions in height and width or as Y, X pairs.

  The padding size (equal in Y, X)

• optional uint32 pad_h = 9

  The padding height

• optional uint32 pad_w = 10

  The padding width

• optional uint32 kernel_size = 2

  The kernel size (square)

• optional uint32 kernel_h = 5

  The kernel height

• optional uint32 kernel_w = 6

  The kernel width

• optional uint32 stride = 3

  The stride (equal in Y, X)

• optional uint32 stride_h = 7

  The stride height

• optional uint32 stride_w = 8

  The stride width

• optional PoolingParameter.Engine engine = 11

• optional bool global_pooling = 12

  If global_pooling then it will pool over the size of the bottom by doing kernel_h = bottom->height and kernel_w = bottom->width

Used in: PoolingParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: PoolingParameter

• MAX = 0

• AVE = 1

Used in: LayerParameter, V1LayerParameter

• optional float power = 1

  PowerLayer computes outputs y = (shift + scale * x) ^ power.

• optional float scale = 2

• optional float shift = 3

Message that store parameters used by PriorBoxLayer

Used in: LayerParameter

• repeated float min_size = 1

  Minimum box size (in pixels). Required!

• repeated float max_size = 2

  Maximum box size (in pixels). Required!

• repeated float aspect_ratio = 3

  Various of aspect ratios. Duplicate ratios will be ignored. If none is provided, we use default ratio 1.

• optional bool flip = 4

  If true, will flip each aspect ratio. For example, if there is aspect ratio "r", we will generate aspect ratio "1.0/r" as well.

• optional bool clip = 5

  If true, will clip the prior so that it is within [0, 1]

• repeated float variance = 6

  Variance for adjusting the prior bboxes.

• optional uint32 img_size = 7

  By default, we calculate img_height, img_width, step_x, step_y based on bottom[0] (feat) and bottom[1] (img). Unless these values are explicitely provided. Explicitly provide the img_size.

• optional uint32 img_h = 8

  Either img_size or img_h/img_w should be specified; not both.

• optional uint32 img_w = 9

• optional float step = 10

  Explicitly provide the step size.

• optional float step_h = 11

  Either step or step_h/step_w should be specified; not both.

• optional float step_w = 12

• optional float offset = 13

  Offset to the top left corner of each cell.

Encode/decode type.

Used in: DetectionOutputParameter

• CORNER = 1

• CENTER_SIZE = 2

• CORNER_SIZE = 3

Used in: LayerParameter

• optional uint32 feat_stride = 1

• optional uint32 base_size = 2

• optional uint32 min_size = 3

• repeated float ratio = 4

• repeated float scale = 5

• optional uint32 pre_nms_topn = 6

• optional uint32 post_nms_topn = 7

• optional float nms_thresh = 8

Used in: LayerParameter

• optional uint32 pooled_h = 1

  Pad, kernel size, and stride are all given as a single value for equal dimensions in height and width or as Y, X pairs.

  The pooled output height

• optional uint32 pooled_w = 2

  The pooled output width

• optional float spatial_scale = 3

  Multiplicative spatial scale factor to translate ROI coords from their input scale to the scale used when pooling

Message that stores parameters used by ReLULayer

Used in: LayerParameter, V1LayerParameter

• optional float negative_slope = 1

  Allow non-zero slope for negative inputs to speed up optimization Described in: Maas, A. L., Hannun, A. Y., & Ng, A. Y. (2013). Rectifier nonlinearities improve neural network acoustic models. In ICML Workshop on Deep Learning for Audio, Speech, and Language Processing.

• optional ReLUParameter.Engine engine = 2

Used in: ReLUParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Message that stores parameters used by ReductionLayer

Used in: LayerParameter

• optional ReductionParameter.ReductionOp operation = 1

  reduction operation

• optional int32 axis = 2

  The first axis to reduce to a scalar -- may be negative to index from the end (e.g., -1 for the last axis). (Currently, only reduction along ALL "tail" axes is supported; reduction of axis M through N, where N < num_axes - 1, is unsupported.) Suppose we have an n-axis bottom Blob with shape: (d0, d1, d2, ..., d(m-1), dm, d(m+1), ..., d(n-1)). If axis == m, the output Blob will have shape (d0, d1, d2, ..., d(m-1)), and the ReductionOp operation is performed (d0 * d1 * d2 * ... * d(m-1)) times, each including (dm * d(m+1) * ... * d(n-1)) individual data. If axis == 0 (the default), the output Blob always has the empty shape (count 1), performing reduction across the entire input -- often useful for creating new loss functions.

• optional float coeff = 3

  coefficient for output

Used in: ReductionParameter

• SUM = 1

• ASUM = 2

• SUMSQ = 3

• MEAN = 4

Used in: LayerParameter

• optional BlobShape shape = 1

  Specify the output dimensions. If some of the dimensions are set to 0, the corresponding dimension from the bottom layer is used (unchanged). Exactly one dimension may be set to -1, in which case its value is inferred from the count of the bottom blob and the remaining dimensions. For example, suppose we want to reshape a 2D blob "input" with shape 2 x 8: layer { type: "Reshape" bottom: "input" top: "output" reshape_param { ... } } If "input" is 2D with shape 2 x 8, then the following reshape_param specifications are all equivalent, producing a 3D blob "output" with shape 2 x 2 x 4: reshape_param { shape { dim: 2 dim: 2 dim: 4 } } reshape_param { shape { dim: 0 dim: 2 dim: 4 } } reshape_param { shape { dim: 0 dim: 2 dim: -1 } } reshape_param { shape { dim: 0 dim:-1 dim: 4 } }

• optional int32 axis = 2

  axis and num_axes control the portion of the bottom blob's shape that are replaced by (included in) the reshape. By default (axis == 0 and num_axes == -1), the entire bottom blob shape is included in the reshape, and hence the shape field must specify the entire output shape. axis may be non-zero to retain some portion of the beginning of the input shape (and may be negative to index from the end; e.g., -1 to begin the reshape after the last axis, including nothing in the reshape, -2 to include only the last axis, etc.). For example, suppose "input" is a 2D blob with shape 2 x 8. Then the following ReshapeLayer specifications are all equivalent, producing a blob "output" with shape 2 x 2 x 4: reshape_param { shape { dim: 2 dim: 2 dim: 4 } } reshape_param { shape { dim: 2 dim: 4 } axis: 1 } reshape_param { shape { dim: 2 dim: 4 } axis: -3 } num_axes specifies the extent of the reshape. If num_axes >= 0 (and axis >= 0), the reshape will be performed only on input axes in the range [axis, axis+num_axes]. num_axes may also be -1, the default, to include all remaining axes (starting from axis). For example, suppose "input" is a 2D blob with shape 2 x 8. Then the following ReshapeLayer specifications are equivalent, producing a blob "output" with shape 1 x 2 x 8. reshape_param { shape { dim: 1 dim: 2 dim: 8 } } reshape_param { shape { dim: 1 dim: 2 } num_axes: 1 } reshape_param { shape { dim: 1 } num_axes: 0 } On the other hand, these would produce output blob shape 2 x 1 x 8: reshape_param { shape { dim: 2 dim: 1 dim: 8 } } reshape_param { shape { dim: 1 } axis: 1 num_axes: 0 }

• optional int32 num_axes = 3

Used in: LayerParameter

• optional uint32 pyramid_height = 1

• optional SPPParameter.PoolMethod pool = 2

  The pooling method

• optional SPPParameter.Engine engine = 6

Used in: SPPParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: SPPParameter

• MAX = 0

• AVE = 1

Used in: LayerParameter

• optional int32 axis = 1

  The first axis of bottom[0] (the first input Blob) along which to apply bottom[1] (the second input Blob). May be negative to index from the end (e.g., -1 for the last axis). For example, if bottom[0] is 4D with shape 100x3x40x60, the output top[0] will have the same shape, and bottom[1] may have any of the following shapes (for the given value of axis): (axis == 0 == -4) 100; 100x3; 100x3x40; 100x3x40x60 (axis == 1 == -3) 3; 3x40; 3x40x60 (axis == 2 == -2) 40; 40x60 (axis == 3 == -1) 60 Furthermore, bottom[1] may have the empty shape (regardless of the value of "axis") -- a scalar multiplier.

• optional int32 num_axes = 2

  (num_axes is ignored unless just one bottom is given and the scale is a learned parameter of the layer. Otherwise, num_axes is determined by the number of axes by the second bottom.) The number of axes of the input (bottom[0]) covered by the scale parameter, or -1 to cover all axes of bottom[0] starting from `axis`. Set num_axes := 0, to multiply with a zero-axis Blob: a scalar.

• optional FillerParameter filler = 3

  (filler is ignored unless just one bottom is given and the scale is a learned parameter of the layer.) The initialization for the learned scale parameter. Default is the unit (1) initialization, resulting in the ScaleLayer initially performing the identity operation.

• optional bool bias_term = 4

  Whether to also learn a bias (equivalent to a ScaleLayer+BiasLayer, but may be more efficient). Initialized with bias_filler (defaults to 0).

• optional FillerParameter bias_filler = 5

Used in: LayerParameter

• optional uint32 group = 1

  The number of group

Used in: LayerParameter, V1LayerParameter

• optional SigmoidParameter.Engine engine = 1

Used in: SigmoidParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: LayerParameter, V1LayerParameter

• optional int32 axis = 3

  The axis along which to slice -- may be negative to index from the end (e.g., -1 for the last axis). By default, SliceLayer concatenates blobs along the "channels" axis (1).

• repeated uint32 slice_point = 2

• optional uint32 slice_dim = 1

  DEPRECATED: alias for "axis" -- does not support negative indexing.

Message that stores parameters used by SoftmaxLayer, SoftmaxWithLossLayer

Used in: LayerParameter, V1LayerParameter

• optional SoftmaxParameter.Engine engine = 1

• optional int32 axis = 2

  The axis along which to perform the softmax -- may be negative to index from the end (e.g., -1 for the last axis). Any other axes will be evaluated as independent softmaxes.

Used in: SoftmaxParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Used in: LayerParameter, V1LayerParameter

• optional TanHParameter.Engine engine = 1

Used in: TanHParameter

• DEFAULT = 0

• CAFFE = 1

• CUDNN = 2

Message that stores parameters used by ThresholdLayer

Used in: LayerParameter, V1LayerParameter

• optional float threshold = 1

  Strictly positive values

Message that stores parameters used by TileLayer

Used in: LayerParameter

• optional int32 axis = 1

  The index of the axis to tile.

• optional int32 tiles = 2

  The number of copies (tiles) of the blob to output.

DEPRECATED: use LayerParameter.

Used in: NetParameter

• repeated string bottom = 2

• repeated string top = 3

• optional string name = 4

• repeated NetStateRule include = 32

• repeated NetStateRule exclude = 33

• optional V1LayerParameter.LayerType type = 5

• repeated BlobProto blobs = 6

• repeated string param = 1001

• repeated V1LayerParameter.DimCheckMode blob_share_mode = 1002

• repeated float blobs_lr = 7

• repeated float weight_decay = 8

• repeated float loss_weight = 35

• optional ConcatParameter concat_param = 9

• optional ConvolutionParameter convolution_param = 10

• optional DropoutParameter dropout_param = 12

• optional EltwiseParameter eltwise_param = 24

• optional ExpParameter exp_param = 41

• optional InnerProductParameter inner_product_param = 17

• optional LRNParameter lrn_param = 18

• optional MVNParameter mvn_param = 34

• optional PoolingParameter pooling_param = 19

• optional PowerParameter power_param = 21

• optional ReLUParameter relu_param = 30

• optional SigmoidParameter sigmoid_param = 38

• optional SoftmaxParameter softmax_param = 39

• optional SliceParameter slice_param = 31

• optional TanHParameter tanh_param = 37

• optional ThresholdParameter threshold_param = 25

• optional LossParameter loss_param = 42

Used in: V1LayerParameter

• STRICT = 0

• PERMISSIVE = 1

Used in: V1LayerParameter

• NONE = 0

• ABSVAL = 35

• BNLL = 2

• CONCAT = 3

• CONVOLUTION = 4

• DECONVOLUTION = 39

• DROPOUT = 6

• ELTWISE = 25

• EXP = 38

• FLATTEN = 8

• INNER_PRODUCT = 14

• LRN = 15

• MVN = 34

• POOLING = 17

• POWER = 26

• RELU = 18

• SIGMOID = 19

• SOFTMAX = 20

• SPLIT = 22

• SLICE = 33

• TANH = 23

• THRESHOLD = 31