View Source Bumblebee.Text.Bart (Bumblebee v0.6.0)
BART model family.
Architectures
:base
- plain BART without any head on top:for_causal_language_modeling
- BART with a language modeling head. The head returns logits for each token in the original sequence:for_conditional_generation
- BART with a language modeling head. The head returns logits for each token in the original sequence:for_sequence_classification
- BART with a sequence classification head. The head returns logits corresponding to possible classes:for_question_answering
- BART with a span classification head. The head returns logits for the span start and end positions
Inputs
"input_ids"
-{batch_size, sequence_length}
Indices of input sequence tokens in the vocabulary.
"attention_mask"
-{batch_size, sequence_length}
Mask indicating which tokens to attend to. This is used to ignore padding tokens, which are added when processing a batch of sequences with different length.
"position_ids"
-{batch_size, sequence_length}
Indices of positions of each input sequence tokens in the position embeddings.
"attention_head_mask"
-{encoder_num_blocks, encoder_num_attention_heads}
Mask to nullify selected heads of the self-attention blocks in the encoder.
"input_embeddings"
-{batch_size, sequence_length, hidden_size}
Embedded representation of
"input_ids"
, which can be specified for more control over how"input_ids"
are embedded than the model's internal embedding lookup. If"input_embeddings"
are present, then"input_ids"
will be ignored."decoder_input_ids"
-{batch_size, target_sequence_length}
Indices of decoder input sequence tokens in the vocabulary. If not present and
"input_ids"
is, it will be generated by shifting each token in"input_ids"
to the right once."decoder_attention_mask"
-{batch_size, target_sequence_length}
Mask indicating which decoder tokens to attend to. This is used to ignore padding tokens, which are added when processing a batch of sequences with different length.
"decoder_position_ids"
-{batch_size, target_sequence_length}
Indices of positions of each decoder input sequence tokens in the position embeddings.
"decoder_attention_head_mask"
-{decoder_num_blocks, decoder_num_attention_heads}
Mask to nullify selected heads of the self-attention blocks in the decoder.
"decoder_input_embeddings"
-{batch_size, sequence_length, hidden_size}
Embedded representation of
"decoder_input_ids"
, which can be specified for more control over how"decoder_input_ids"
are embedded than the model's internal embedding lookup. If"decoder_input_embeddings"
are present, then"decoder_input_ids"
will be ignored."encoder_hidden_state"
-{batch_size, sequence_length, hidden_size}
Last hidden state output from the encoder. This hidden state is used in cross-attention blocks in the decoder. If specified, the model will skip the encoding process and use this value directly for cross-attentions in the decoder.
"cross_attention_head_mask"
-{decoder_num_blocks, decoder_num_attention_heads}
Mask to nullify selected heads of the cross-attention blocks in the decoder with shape.
"cache"
A container with cached layer results used to speed up sequential decoding (autoregression). With cache, certain hidden states are taken from the cache, rather than recomputed on every decoding pass. The cache should be treated as opaque and initialized with
Bumblebee.Text.Generation.init_cache/4
.
Exceptions
The :for_causal_language_modeling
model is just the decoder part and
accepts the following inputs instead: "input_ids"
, "attention_mask"
,
"position_ids"
, "attention_head_mask"
, "input_embeddings"
, "encoder_hidden_state"
,
"encoder_attention_mask"
, "cross_attention_head_mask"
, "cache"
.
Global layer options
:output_hidden_states
- whentrue
, the model output includes all hidden states:output_attentions
- whentrue
, the model output includes all attention weights
Configuration
:vocab_size
- the vocabulary size of the token embedding. This corresponds to the number of distinct tokens that can be represented in model input and output . Defaults to50265
:max_positions
- the vocabulary size of the position embedding. This corresponds to the maximum sequence length that this model can process. Typically this is set to a large value just in case, such as 512, 1024 or 2048 . Defaults to1024
:hidden_size
- the dimensionality of hidden layers. Defaults to1024
:encoder_num_blocks
- the number of Transformer blocks in the encoder. Defaults to12
:decoder_num_blocks
- the number of Transformer blocks in the decoder. Defaults to12
:encoder_num_attention_heads
- the number of attention heads for each attention layer in the encoder. Defaults to16
:decoder_num_attention_heads
- the number of attention heads for each attention layer in the decoder. Defaults to16
:scale_embedding
- whether to scale embeddings by dividing by the square root of:hidden_size
. Defaults tofalse
:activation
- the activation function. Defaults to:gelu
:dropout_rate
- the dropout rate for encoder and decoder. Defaults to0.1
:attention_dropout_rate
- the dropout rate for attention weights. Defaults to0.0
:activation_dropout_rate
- the dropout rate for activations inside fully connected layers. Defaults to0.0
:classifier_dropout_rate
- the dropout rate for the classification head. Defaults to0.0
:initializer_scale
- the standard deviation of the normal initializer used for initializing kernel parameters. Defaults to0.02
:num_labels
- the number of labels to use in the last layer for the classification task. Defaults to2
:id_to_label
- a map from class index to label. Defaults to%{}