- ot = f(Xt*(Wo^T) + Ht-1*(Ro^T) + Po (.) Ct + Wbo + Rbo)
- Ht = ot (.) h(Ct)
AttentionWrapp Notations:
`lstm()' - wrapped inner cell.
Ht, Ct = lstm(concat(Xt, ATTNt-1), Ct-1)
`am()` - attention mechanism the wrapper used.
CONTEXTt, ALIGNt = am(Ht, ALIGNt-1)
`AW` - attention layer weights, optional.
`ATTN` - attention state, initial is zero. If `AW` provided, it is the output of the attention layer,
ATTNt = concat(Ht, CONTEXTt) * AW
otherwise,
ATTNt = CONTEXTt
RNN layer output:
`Y` - if needed is the sequence of Ht from lstm cell.
`Y_h` - is the last valid H from lstm cell.
`Y_c` - is the last valid C from lstm cell.
#### Version
This version of the operator has been available since version 1 of the 'com.microsoft' operator set.
#### Attributes
<dl>
<dt><tt>activation_alpha</tt> : list of floats</dt>
<dd>Optional scaling values used by some activation functions. The values are consumed in the order of activation functions, for example (f, g, h) in LSTM. Default values are the same as of corresponding ONNX operators.For example with LeakyRelu, the default alpha is 0.01.</dd>
<dt><tt>activation_beta</tt> : list of floats</dt>
<dd>Optional scaling values used by some activation functions. The values are consumed in the order of activation functions, for example (f, g, h) in LSTM. Default values are the same as of corresponding ONNX operators.</dd>
<dt><tt>activations</tt> : list of strings</dt>
<dd>A list of 3 (or 6 if bidirectional) activation functions for input, output, forget, cell, and hidden. The activation functions must be one of the activation functions specified above. Optional: See the equations for default if not specified.</dd>
<dt><tt>clip</tt> : float</dt>
<dd>Cell clip threshold. Clipping bounds the elements of a tensor in the range of [-threshold, +threshold] and is applied to the input of activations. No clip if not specified.</dd>
<dt><tt>direction</tt> : string</dt>
<dd>Specify if the RNN is forward, reverse, or bidirectional. Must be one of forward (default), reverse, or bidirectional.</dd>
<dt><tt>hidden_size</tt> : int</dt>
<dd>Number of neurons in the hidden layer.</dd>
<dt><tt>input_forget</tt> : int</dt>
<dd>Couple the input and forget gates if 1, default 0.</dd>
<dd>The input sequences packed (and potentially padded) into one 3-D tensor with the shape of `[seq_length, batch_size, input_size]`</dd>
<dt><tt>W</tt> : T</dt>
<dd>The weight tensor for the gates. Concatenation of `W[iofc]` and `WB[iofc]` (if bidirectional) along dimension 0. The tensor has shape `[num_directions, 4*hidden_size, input_size]`.</dd>
<dt><tt>R</tt> : T</dt>
<dd>The recurrence weight tensor. Concatenation of `R[iofc]` and `RB[iofc]` (if bidirectional) along dimension 0. This tensor has shape `[num_directions, 4*hidden_size, hidden_size]`.</dd>
<dt><tt>B</tt> (optional) : T</dt>
<dd>The bias tensor for input gate. Concatenation of `[Wb[iofc], Rb[iofc]]`, and `[WBb[iofc], RBb[iofc]]` (if bidirectional) along dimension 0. This tensor has shape `[num_directions, 8*hidden_size]`. Optional: If not specified - assumed to be 0.</dd>
<dt><tt>sequence_lens</tt> (optional) : T1</dt>
<dd>Optional tensor specifying lengths of the sequences in a batch. If not specified - assumed all sequences in the batch to have length `seq_length`. It has shape `[batch_size]`</dd>
<dt><tt>initial_h</tt> (optional) : T</dt>
<dd>Optional initial value of the hidden. If not specified - assumed to be 0. It has shape `[num_directions, batch_size, hidden_size]`.</dd>
<dt><tt>initial_c</tt> (optional) : T</dt>
<dd>Optional initial value of the cell. If not specified - assumed to be 0. It has shape `[num_directions, batch_size, hidden_size]`.</dd>
<dt><tt>P</tt> (optional) : T</dt>
<dd>The weight tensor for peepholes. Concatenation of `P[iof]` and `PB[iof]` (if bidirectional) along dimension 0. It has shape `[num_directions, 3*hidde_size]`. Optional: If not specified - assumed to be 0.</dd>
<dt><tt>QW</tt> (optional) : T</dt>
<dd>The weight tensor of the query layer in the attention mechanism. Should be of shape `[num_directions, am_query_depth(hidden_size of lstm), am_attn_size]`</dd>
<dt><tt>MW</tt> (optional) : T</dt>
<dd>The weight tensor of the memory layer in the attention mechanism. Should be of shape `[num_directions, memory_depth, am_attn_size]`</dd>
<dt><tt>V</tt> (optional) : T</dt>
<dd>The attention_v tensor in the attention mechanism. Should be of shape `[num_directions, am_attn_size]`</dd>
<dt><tt>M</tt> (optional) : T</dt>
<dd>The sequence of the memory (input) for attention mechanism. Should be of `[batch_size, max_memory_step, memory_depth]`</dd>
<dt><tt>memory_seq_lens</tt> (optional) : T1</dt>
<dd>The sequence length of the input memory for the attention mechanism. Should be of `[batch_size]`</dd>
<dt><tt>AW</tt> (optional) : T</dt>
<dd>The weights of attention layer in the attention wrapper. If exists, should be of shape `[num_directions, memory_depth+hidden_size, aw_attn_size]. Please note that attention mechanism context depth is also memory_depth in the attention mechanism.`</dd>
Extracts crops from the input image tensor and resizes them using bilinear sampling or nearest neighbor sampling
(possibly with aspect ratio change) to a common output size specified by crop_height and crop_width.
Returns a tensor with crops from the input image at positions defined at the bounding box locations in boxes.
The cropped boxes are all resized (with bilinear or nearest neighbor interpolation) to
a fixed size = [crop_height, crop_width]. The result is a 4-D tensor [num_boxes, crop_height, crop_width, depth].
The resizing is corner aligned.
#### Version
This version of the operator has been available since version 1 of the 'com.microsoft' operator set.
#### Attributes
<dl>
<dt><tt>extrapolation_value</tt> : float</dt>
<dd>Value used for extrapolation, when applicable. Default is 0.0f. </dd>
<dt><tt>mode</tt> : string</dt>
<dd>The pooling method. Two modes are supported: 'bilinear' and 'nearest'. Default is 'bilinear'.</dd>
</dl>
#### Inputs
<dl>
<dt><tt>X</tt> : T1</dt>
<dd>Input data tensor from the previous operator; 4-D feature map of shape (N, C, H, W), where N is the batch size, C is the number of channels, and H and W are the height and the width of the data.</dd>
<dt><tt>rois</tt> : T1</dt>
<dd>RoIs (Regions of Interest) to pool over; rois is 2-D input of shape (num_rois, 4) given as [[y1, x1, y2, x2], ...]. The RoIs' coordinates are normalized in the coordinate system of the input image. Each coordinate set has a 1:1 correspondence with the 'batch_indices' input.</dd>
<dt><tt>batch_indices</tt> : T2</dt>
<dd>1-D tensor of shape (num_rois,) with each element denoting the index of the corresponding image in the batch.</dd>
<dt><tt>crop_size</tt> : T2</dt>
<dd>1-D tensor of 2 elements: [crop_height, crop_width]. All cropped image patches are resized to this size. Both crop_height and crop_width need to be positive.</dd>
</dl>
#### Outputs
<dl>
<dt><tt>Y</tt> : T1</dt>
<dd>RoI pooled output, 4-D tensor of shape (num_rois, C, crop_height, crop_width). The r-th batch element Y[r-1] is a pooled feature map corresponding to the r-th RoI X[r-1].</dd>
The underlying implementation is MurmurHash3_x86_32 generating low latency 32bits hash suitable for implementing lookup tables, Bloom filters, count min sketch or feature hashing.
#### Version
This version of the operator has been available since version 1 of the 'com.microsoft' operator set.
#### Attributes
<dl>
<dt><tt>positive</tt> : int</dt>
<dd>If value is 1, output type is uint32_t, else int32_t. Default value is 1.</dd>
<dt><tt>seed</tt> : int</dt>
<dd>Seed for the hashing algorithm, unsigned 32-bit integer, default to 0.</dd>
Insert 0 pads to the beginning of the second dimension.
data = [
[1.0, 1.2],
[2.3, 3.4],
[4.5, 5.7],
]
pads = [0, 2, 0, 0]
output = [
[
[0.0, 0.0, 1.0, 1.2],
[0.0, 0.0, 2.3, 3.4],
[0.0, 0.0, 4.5, 5.7],
],
]
#### Version
This version of the operator has been available since version 1 of the 'com.microsoft' operator set.
#### Attributes
<dl>
<dt><tt>mode</tt> : string</dt>
<dd>Three modes: `constant`(default) - pads with a given constant value, `reflect` - pads with the reflection of the vector mirrored on the first and last values of the vector along each axis, `edge` - pads with the edge values of array</dd>
</dl>
#### Inputs (2 - 3)
<dl>
<dt><tt>data</tt> : T</dt>
<dd>Input tensor.</dd>
<dt><tt>pads</tt> : tensor(int64)</dt>
<dd>Tensor of integers indicating the number of padding elements to add or remove (if negative) at the beginning and end of each axis. For 2D input tensor, it is the number of pixels. `pads` should be a 1D tensor of shape [2 * input_rank] or a 2D tensor of shape [1, 2 * input_rank]. `pads` format (1D example) should be as follow [x1_begin, x2_begin,...,x1_end, x2_end,...], where xi_begin is the number of pixels added at the beginning of axis `i` and xi_end, the number of pixels added at the end of axis `i`.</dd>
<dt><tt>value</tt> (optional) : T</dt>
<dd>(Optional) A scalar or rank 1 tensor containing a single value to be filled if the mode chosen is `constant` (by default it is 0.0).</dd>
<dd>Constrain output data type to 32-bit integer tensor.T2 must be tensor(uint32) when T1 is tensor(uint8),or must be tensor(int32) when T1 is tensor(int8).</dd>
This version of the operator has been available since version 1 of the 'com.microsoft' operator set.
#### Attributes
<dl>
<dt><tt>mark</tt> : int (required)</dt>
<dd>Boolean whether to mark the beginning/end character with start of text character (0x02)/end of text character (0x03).</dd>
<dt><tt>mincharnum</tt> : int (required)</dt>
<dd>Minimum number of characters allowed in the output. For example, if mincharnum is 2, tokens such as "A" and "B" would be ignored</dd>
<dt><tt>pad_value</tt> : string (required)</dt>
<dd>The string used to pad output tensors when the tokens extracted doesn't match the maximum number of tokens found. If start/end markers are needed, padding will appear outside the markers.</dd>
<dd>an optional list of strings attribute that contains a list of separators - regular expressions to match separators Two consecutive segments in X connected by a separator would be divided into two tokens. For example, if the input is "Hello World!" and this attribute contains only one space character, the corresponding output would be ["Hello", "World!"]. To achieve character-level tokenization, one should set the 'separators' to [""], which contains an empty string.</dd>
<dd>An optional string. Token's regular expression in basic POSIX format (http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap09.html#tag_09_03). If set, tokenizer may produce tokens matching the specified pattern. Note that one and only of 'tokenexp' and 'separators' should be set.</dd>
The WordConvEmbedding takes in a batch of sequence words and embed each word to a vector.
#### Version
This version of the operator has been available since version 1 of the 'com.microsoft' operator set.
#### Attributes
<dl>
<dt><tt>char_embedding_size</tt> : int</dt>
<dd>Integer representing the embedding vector size for each char.If not provide, use the char embedding size of embedding vector.</dd>
<dt><tt>conv_window_size</tt> : int</dt>
<dd>This operator applies convolution to word from left to right with window equal to conv_window_size and stride to 1.Take word 'example' for example, with conv_window_size equal to 2, conv is applied to [ex],[xa], [am], [mp]...If not provide, use the first dimension of conv kernal shape.</dd>
<dt><tt>embedding_size</tt> : int</dt>
<dd>Integer representing the embedding vector size for each word.If not provide, use the fileter size of conv weight</dd>
