Struct BartGenerator

pub struct BartGenerator { /* private fields */ }

Language generation model based on the Bart architecture

Implementations

impl BartGenerator

pub fn new( generate_config: GenerateConfig, ) -> Result<BartGenerator, RustBertError>

Build a new BartGenerator

Arguments

• vocab_path - Path to the model vocabulary, expected to have a structure following the Transformers library convention
• merges_path - Path to the bpe merges, expected to have a structure following the Transformers library convention
• config_path - Path to the model configuration, expected to have a structure following the Transformers library convention
• weights_path - Path to the model weight files. These need to be converted form the .bin to .ot format using the utility script provided.
• device - Device to run the model on, e.g. Device::Cpu or Device::Cuda(0)

Example

use rust_bert::bart::BartGenerator;
use rust_bert::pipelines::generation_utils::GenerateConfig;
let device = Device::cuda_if_available();
let generate_config = GenerateConfig {
    max_length: Some(30),
    do_sample: true,
    num_beams: 5,
    temperature: 1.1,
    num_return_sequences: 3,
    ..Default::default()
};
let bart_generator = BartGenerator::new(generate_config)?;

pub fn new_with_tokenizer( generate_config: GenerateConfig, tokenizer: TokenizerOption, ) -> Result<BartGenerator, RustBertError>

Trait Implementations

impl LanguageGenerator for BartGenerator

fn generate<S>( &self, prompt_texts: Option<&[S]>, generate_options: Option<GenerateOptions<'_>>, ) -> Result<Vec<GeneratedTextOutput>, RustBertError>

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

Generate text based on a vector of promp texts.

fn generate_indices<S>( &self, prompt_texts: Option<&[S]>, generate_options: Option<GenerateOptions<'_>>, ) -> Result<Vec<GeneratedIndicesOutput>, RustBertError>

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

Generate token indices without decoding (useful for token-level operations before returning final text or as validation step during training).

fn generate_from_ids_and_past( &self, input_ids: Tensor, attention_mask: Option<Tensor>, generate_options: Option<GenerateOptions<'_>>, ) -> Result<Vec<GeneratedIndicesOutput>, RustBertError>

Generate token indices given a list of indices (useful when the input has been pre-tokenized). Returns a list of output tokens that need to be decoded using a tokenizer.

fn get_tokenizer(&self) -> &TokenizerOption

Returns a reference to the text generator’s tokenizer

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

fn half(&mut self) -> Result<(), RustBertError>

fn float(&mut self) -> Result<(), RustBertError>

fn set_device(&mut self, device: Device) -> Result<(), RustBertError>