class allennlp.models.decomposable_attention.DecomposableAttention(vocab:, text_field_embedder: allennlp.modules.text_field_embedders.text_field_embedder.TextFieldEmbedder, attend_feedforward: allennlp.modules.feedforward.FeedForward, similarity_function: allennlp.modules.similarity_functions.similarity_function.SimilarityFunction, compare_feedforward: allennlp.modules.feedforward.FeedForward, aggregate_feedforward: allennlp.modules.feedforward.FeedForward, premise_encoder: Optional[allennlp.modules.seq2seq_encoders.seq2seq_encoder.Seq2SeqEncoder] = None, hypothesis_encoder: Optional[allennlp.modules.seq2seq_encoders.seq2seq_encoder.Seq2SeqEncoder] = None, initializer: allennlp.nn.initializers.InitializerApplicator = <allennlp.nn.initializers.InitializerApplicator object>, regularizer: Optional[allennlp.nn.regularizers.regularizer_applicator.RegularizerApplicator] = None)[source]

Bases: allennlp.models.model.Model

This Model implements the Decomposable Attention model described in “A Decomposable Attention Model for Natural Language Inference” by Parikh et al., 2016, with some optional enhancements before the decomposable attention actually happens. Parikh’s original model allowed for computing an “intra-sentence” attention before doing the decomposable entailment step. We generalize this to any Seq2SeqEncoder that can be applied to the premise and/or the hypothesis before computing entailment.

The basic outline of this model is to get an embedded representation of each word in the premise and hypothesis, align words between the two, compare the aligned phrases, and make a final entailment decision based on this aggregated comparison. Each step in this process uses a feedforward network to modify the representation.


Used to embed the premise and hypothesis TextFields we get as input to the model.


This feedforward network is applied to the encoded sentence representations before the similarity matrix is computed between words in the premise and words in the hypothesis.


This is the similarity function used when computing the similarity matrix between words in the premise and words in the hypothesis.


This feedforward network is applied to the aligned premise and hypothesis representations, individually.


This final feedforward network is applied to the concatenated, summed result of the compare_feedforward network, and its output is used as the entailment class logits.

premise_encoderSeq2SeqEncoder, optional (default=``None``)

After embedding the premise, we can optionally apply an encoder. If this is None, we will do nothing.

hypothesis_encoderSeq2SeqEncoder, optional (default=``None``)

After embedding the hypothesis, we can optionally apply an encoder. If this is None, we will use the premise_encoder for the encoding (doing nothing if premise_encoder is also None).

initializerInitializerApplicator, optional (default=``InitializerApplicator()``)

Used to initialize the model parameters.

regularizerRegularizerApplicator, optional (default=``None``)

If provided, will be used to calculate the regularization penalty during training.

forward(self, premise: Dict[str, torch.LongTensor], hypothesis: Dict[str, torch.LongTensor], label: torch.IntTensor = None, metadata: List[Dict[str, Any]] = None) → Dict[str, torch.Tensor][source]
premiseDict[str, torch.LongTensor]

From a TextField

hypothesisDict[str, torch.LongTensor]

From a TextField

labeltorch.IntTensor, optional, (default = None)

From a LabelField

metadataList[Dict[str, Any]], optional, (default = None)

Metadata containing the original tokenization of the premise and hypothesis with ‘premise_tokens’ and ‘hypothesis_tokens’ keys respectively.

An output dictionary consisting of:

A tensor of shape (batch_size, num_labels) representing unnormalised log probabilities of the entailment label.


A tensor of shape (batch_size, num_labels) representing probabilities of the entailment label.

losstorch.FloatTensor, optional

A scalar loss to be optimised.

get_metrics(self, reset: bool = False) → Dict[str, float][source]

Returns a dictionary of metrics. This method will be called by in order to compute and use model metrics for early stopping and model serialization. We return an empty dictionary here rather than raising as it is not required to implement metrics for a new model. A boolean reset parameter is passed, as frequently a metric accumulator will have some state which should be reset between epochs. This is also compatible with Metrics should be populated during the call to ``forward`, with the Metric handling the accumulation of the metric until this method is called.