TrEncoderLayer
The TrEncoderLayer class consists of a Multi-Head Self-Attention and a Feed-Forward Neural Network (FFNN) sub-layers.
Value parameters
- f
-
the activation function family (used by alinear1)
- heads
-
the number of attention heads
- n_mod
-
the size of the output (dimensionality of the model, d_model)
- n_v
-
the size of the value vectors
- n_var
-
the size of the input vector x_t (number of variables)
- n_z
-
the size of the hidden layer in the Feed-Forward Neural Network
- norm_eps
-
a small values used in normalization to avoid divide by zero
- norm_first
-
whether layer normalization should be done first (see apply method)
- p_drop
-
the probability of setting an element to zero in a dropout layer
Attributes
- See also
-
pytorch.org/docs/stable/generated/torch.nn.TransformerEncoderLayer.html#torch.nn.TransformerEncoderLayer
- Graph
-
- Supertypes
Members list
Value members
Concrete methods
Forward pass: Compute this encoder layer's result z by using Multi-Head Self-Attention followed by a Feed-Forward Neural Network.
Forward pass: Compute this encoder layer's result z by using Multi-Head Self-Attention followed by a Feed-Forward Neural Network.
Value parameters
- x
-
the input matrix
Attributes
Compute the Feed-forward Neural Network result.
Compute the Feed-forward Neural Network result.
Value parameters
- x
-
the input matrix
Attributes
Inherited methods
Compute a Self-Attention Weight Matrix from the given query (Q), key (K) and value (V).
Compute a Self-Attention Weight Matrix from the given query (Q), key (K) and value (V).
Value parameters
- k
-
the key matrix K
- q
-
the query matrix Q (q_t over all time)
- v
-
the value matrix V
Attributes
- Inherited from:
- Attention
Compute a Multi-Head, Self-Attention Weight Matrix by taking attention for each head and concatenating them; finally multiplying by the overall weight matrix w_o. The operator ++^ concatenates matrices column-wise.
Compute a Multi-Head, Self-Attention Weight Matrix by taking attention for each head and concatenating them; finally multiplying by the overall weight matrix w_o. The operator ++^ concatenates matrices column-wise.
Value parameters
- k
-
the key matrix K
- q
-
the query matrix Q (q_t over all time)
- v
-
the value matrix V
- w_o
-
the overall weight matrix to be applied to concatenated attention
- w_q
-
the weight tensor for query Q (w_q(i) matrix for i-th head)
- w_v
-
the weight tensor for value V (w_v(i) matrix for i-th head)
Attributes
- Inherited from:
- Attention
Compute a Context Vector from the given query at time t (q_t), key (K) and value (V).
Compute a Context Vector from the given query at time t (q_t), key (K) and value (V).
Value parameters
- k
-
the key matrix K
- q_t
-
the query vector at time t (based on input vector x_t)
- v
-
the value matrix V
Attributes
- Inherited from:
- Attention
Compute the Query, Key, Value matrices from the given input and weight matrices.
Compute the Query, Key, Value matrices from the given input and weight matrices.
Value parameters
- w_q
-
the weight matrix for query Q
- w_v
-
the weight matrix for value V
- x
-
the input matrix
Attributes
- Inherited from:
- Attention