Struct SentenceEmbeddingsModel

pub struct SentenceEmbeddingsModel { /* private fields */ }

SentenceEmbeddingsModel to perform sentence embeddings

It is made of the following blocks:

• transformer: Base transformer model
• pooling: Pooling layer
• dense (optional): Linear (feed forward) layer
• normalization (optional): Embeddings normalization

Implementations

impl SentenceEmbeddingsModel

pub fn new(config: SentenceEmbeddingsConfig) -> Result<Self, RustBertError>

Build a new SentenceEmbeddingsModel

Arguments

• config - SentenceEmbeddingsConfig object containing the resource references (model, vocabulary, configuration) and device placement (CPU/GPU)

pub fn new_with_tokenizer( config: SentenceEmbeddingsConfig, tokenizer: TokenizerOption, ) -> Result<Self, RustBertError>

Build a new ONNXCausalGenerator from a GenerateConfig and TokenizerOption.

A tokenizer must be provided by the user and can be customized to use non-default settings.

Arguments

• config - SentenceEmbeddingsConfig object containing the resource references (model, vocabulary, configuration) and device placement (CPU/GPU)
• tokenizer - TokenizerOption tokenizer to use for question answering.

pub fn get_tokenizer(&self) -> &TokenizerOption

Get a reference to the model tokenizer.

pub fn get_tokenizer_mut(&mut self) -> &mut TokenizerOption

Get a mutable reference to the model tokenizer.

pub fn set_tokenizer_truncation( &mut self, truncation_strategy: TruncationStrategy, )

Sets the tokenizer’s truncation strategy

pub fn get_embedding_dim(&self) -> Result<i64, RustBertError>

Return the embedding output dimension

pub fn tokenize<S>(&self, inputs: &[S]) -> SentenceEmbeddingsTokenizerOutput

where S: AsRef<str> + Send + Sync,

Tokenizes the inputs

pub fn encode_as_tensor<S>( &self, inputs: &[S], ) -> Result<SentenceEmbeddingsModelOutput, RustBertError>

where S: AsRef<str> + Send + Sync,

Computes sentence embeddings, outputs Tensor.

pub fn encode<S>(&self, inputs: &[S]) -> Result<Vec<Embedding>, RustBertError>

where S: AsRef<str> + Send + Sync,

Computes sentence embeddings.

pub fn encode_with_attention<S>( &self, inputs: &[S], ) -> Result<(Vec<Embedding>, Vec<AttentionOutput>), RustBertError>

where S: AsRef<str> + Send + Sync,

Computes sentence embeddings, also outputs AttentionOutputs.