|
1 |
| -Layers supporting hierarchical sequence as input |
2 |
| -================================================ |
3 |
| - |
4 |
| -TBD |
| 1 | +########################### |
| 2 | +Layers that Support Hierarchical Sequences as Input |
| 3 | +########################### |
| 4 | + |
| 5 | +.. contents:: |
| 6 | + |
| 7 | +Overview |
| 8 | +==== |
| 9 | + |
| 10 | +A sequence is a common data type in natural language processing tasks. An independent word can be regarded as a non-sequential input or a 0-level sequence. A sentence made up of words is a single-level sequence; a number of sentences make up a paragraph, which is a double-level sequence. |
| 11 | + |
| 12 | +A double-level sequence is a nested sequence where each element is a single-level sequence. This is a very flexible way of organizing data that helps us construct some complex input information. |
| 13 | + |
| 14 | +We can define non-sequences, single-level sequences, and double-level sequences at the following levels. |
| 15 | + |
| 16 | ++ 0-level sequence: an independent element. Its type can be any input data type supported by PaddlePaddle; |
| 17 | ++ Single-level sequence: multiple elements arranged in a row; each element is a 0-level sequence. The order of elements is an important input information; |
| 18 | ++ Double-level sequence: multiple elements arranged in a row; each element is a single-layer sequence called a subseq of a double-level sequence, and each element of the subseq is a 0-level sequence. |
| 19 | + |
| 20 | +In PaddlePaddle, the following layers accept double-layer sequences as input and perform corresponding calculations. |
| 21 | + |
| 22 | +`pooling` |
| 23 | +======== |
| 24 | + |
| 25 | +The use of pooling is as follows: |
| 26 | + |
| 27 | +.. code-block:: bash |
| 28 | + |
| 29 | + Seq_pool = pooling(input=layer, |
| 30 | + Pooling_type=pooling.Max(), |
| 31 | + Agg_level=AggregateLevel.TO_SEQUENCE) |
| 32 | + |
| 33 | +- `pooling_type` currently supports two types: pooling.Max() and pooling.Avg(). |
| 34 | + |
| 35 | +- When ʻagg_level=AggregateLevel.TO_NO_SEQUENCE` (default): |
| 36 | + |
| 37 | + - Effect: a double-level sequence input will be converted into a 0-level sequence, and a single-level sequence will be converted into a 0-level sequence |
| 38 | + - Input: a double-level sequence or a single-level sequence |
| 39 | + - Output: a 0-level sequence which is the average (or maximum) of the entire input sequence (single or double) |
| 40 | + |
| 41 | +- When ʻagg_level=AggregateLevel.TO_SEQUENCE`: |
| 42 | + |
| 43 | + - Effect: a double-level sequence will be transformed into a single-level sequence |
| 44 | + - Input: a double-level sequence |
| 45 | + - Output: a single-level sequence where each element of the sequence is the average (or maximum) value of each subseq element of the original double-level sequence. |
| 46 | + |
| 47 | +`last_seq` and `first_seq` |
| 48 | +===================== |
| 49 | + |
| 50 | +An example of using `last_seq` is as follows (usage of `first_seq` is similar). |
| 51 | + |
| 52 | +.. code-block:: bash |
| 53 | + |
| 54 | + Last = last_seq(input=layer, |
| 55 | + Agg_level=AggregateLevel.TO_SEQUENCE) |
| 56 | + |
| 57 | +- When ʻagg_level=AggregateLevel.TO_NO_SEQUENCE` (default): |
| 58 | + |
| 59 | + - Effect: a double-level sequence input will be converted into a 0-level sequence, and a single-level sequence will be converted into a 0-level sequence |
| 60 | + - Input: a double-level sequence or a single-level sequence |
| 61 | + - Output: a 0-level sequence, which is the last or the first element of the input sequence (double or single level). |
| 62 | + |
| 63 | +- When ʻagg_level=AggregateLevel.TO_SEQUENCE`: |
| 64 | + - Effect: a double-level sequence will be transformed into a single-level sequence |
| 65 | + - Input: a double-level sequence |
| 66 | + - Output: a single-layer sequence in which each element is the last (or first) element of each subseq in a double-level sequence. |
| 67 | + |
| 68 | +`expand` |
| 69 | +====== |
| 70 | + |
| 71 | +The use of expand is as follows. |
| 72 | + |
| 73 | +.. code-block:: bash |
| 74 | + |
| 75 | + Ex = expand(input=layer1, |
| 76 | + Expand_as=layer2, |
| 77 | + Expand_level=ExpandLevel.FROM_NO_SEQUENCE) |
| 78 | + |
| 79 | +- When `expand_level=ExpandLevel.FROM_NO_SEQUENCE` (default): |
| 80 | + |
| 81 | + - Effect: a 0-level sequence is extended to a single-level sequence or a double-level sequence |
| 82 | + - Input: layer1 must be a 0-level sequence to be extended; layer2 can be a single-level sequence or a double-level sequence that provides the extended length information |
| 83 | + - Output: a single-level sequence or a double-level sequence; the type of the output sequence and the number of elements contained in the sequence are the same as layer2. If the output is a single-level sequence, each element of the single-level sequence will be a copy of the layer1 element. If the output is a double-level sequence, each element in the double-level sequence will be a copy of the layer1 element |
| 84 | + |
| 85 | +- When `expand_level=ExpandLevel.FROM_SEQUENCE`: |
| 86 | + |
| 87 | + - Effect: a single-level sequence is extended to a double-level sequence |
| 88 | + - Input: layer1 must be a single-level sequence to be extended; layer2 must be a double-level sequence providing extended length information |
| 89 | + - Output: a double-level sequence with the same number of elements as that of layer2. It is required that the number of elements in the single-level sequence be the same as the number of subseq in the double-level sequences. The i-th element of the single-level sequence (the 0-level sequence) is expanded into a single-level sequence that constitutes the i-th subseq of the output, the double-level sequence. |
0 commit comments