nlp-longformer/nlp_longformer.py

import numpy
from typing import NamedTuple
import torch
from transformers import LongformerTokenizer, LongformerModel

from towhee.operator import NNOperator

import warnings
warnings.filterwarnings('ignore')

class NlpLongformer(NNOperator):
    """
    NLP embedding operator that uses the pretrained longformer model gathered by huggingface.
    The Longformer model was presented in Longformer: The Long-Document Transformer by Iz Beltagy,
    Matthew E. Peters, Arman Cohan.
    Ref: https://huggingface.co/docs/transformers/v4.16.2/en/model_doc/longformer#transformers.LongformerConfig

    Args:
        model_name (`str`):
            Which model to use for the embeddings.
    """
    def __init__(self, model_name: str) -> None:
        self.model = LongformerModel.from_pretrained(model_name)
        self.tokenizer = LongformerTokenizer.from_pretrained(model_name)

    def __call__(self, txt: str) -> NamedTuple('Outputs', [('feature_vector', numpy.ndarray)]):
        input_ids = torch.tensor(self.tokenizer.encode(txt)).unsqueeze(0)
        attention_mask = None
        outs = self.model(input_ids, attention_mask=attention_mask, output_hidden_states=True)
        feature_vector = outs[1].squeeze(0)
        Outputs = NamedTuple('Outputs', [('feature_vector', numpy.ndarray)])
        return Outputs(feature_vector.detach().numpy())

    def get_model(self):
        return self.model