Struct FNetForSequenceClassification

pub struct FNetForSequenceClassification { /* private fields */ }

FNet for sequence classification

Base FNet model with a classifier head to perform sentence or document-level classification It is made of the following blocks:

• fnet: Base FNet model
• dropout: Dropout layer before the last linear layer
• classifier: linear layer mapping from hidden to the number of classes to predict

Implementations

impl FNetForSequenceClassification

pub fn new<'p, P>( p: P, config: &FNetConfig, ) -> Result<FNetForSequenceClassification, RustBertError>

where P: Borrow<Path<'p>>,

Build a new FNetForSequenceClassification

Arguments

• p - Variable store path for the root of the FNet model
• config - FNetConfig object defining the model architecture

Example

use rust_bert::fnet::{FNetConfig, FNetForSequenceClassification};
use rust_bert::Config;
use std::path::Path;
use tch::{nn, Device};

let config_path = Path::new("path/to/config.json");
let device = Device::Cpu;
let p = nn::VarStore::new(device);
let config = FNetConfig::from_file(config_path);
let fnet = FNetForSequenceClassification::new(&p.root() / "fnet", &config).unwrap();

pub fn forward_t( &self, input_ids: Option<&Tensor>, token_type_ids: Option<&Tensor>, position_ids: Option<&Tensor>, input_embeddings: Option<&Tensor>, train: bool, ) -> Result<FNetSequenceClassificationOutput, RustBertError>

Forward pass through the model

Arguments

• input_ids - Optional input tensor of shape (batch size, sequence_length). If None, pre-computed embeddings must be provided (see input_embeds)
• token_type_ids - Optional segment id of shape (batch size, sequence_length). Convention is value of 0 for the first sentence (incl. SEP) and 1 for the second sentence. If None set to 0.
• position_ids - Optional position ids of shape (batch size, sequence_length). If None, will be incremented from 0.
• input_embeds - Optional pre-computed input embeddings of shape (batch size, sequence_length, hidden_size). If None, input ids must be provided (see input_ids)
• train - boolean flag to turn on/off the dropout layers in the model. Should be set to false for inference.

Returns

• FNetSequenceClassificationOutput containing:
  • logits - Tensor of shape (batch size, num_classes)
  • all_hidden_states - Option<Vec<Tensor>> of length num_hidden_layers with shape (batch size, sequence_length, hidden_size)

Example

use rust_bert::fnet::{FNetConfig, FNetForSequenceClassification};
let model = FNetForSequenceClassification::new(&vs.root(), &config).unwrap();
let (batch_size, sequence_length) = (64, 128);
let input_tensor = Tensor::rand(&[batch_size, sequence_length], (Int64, device));
let token_type_ids = Tensor::zeros(&[batch_size, sequence_length], (Int64, device));
let position_ids = Tensor::arange(sequence_length, (Int64, device))
    .expand(&[batch_size, sequence_length], true);

let model_output = no_grad(|| {
    model
        .forward_t(
            Some(&input_tensor),
            Some(&token_type_ids),
            Some(&position_ids),
            None,
            false,
        )
        .unwrap()
});