Source code for pyabsa.tasks._Archive.ProteinRegression.models.__plm__.bert

# -*- coding: utf-8 -*-
# file: bert.py
# time: 02/11/2022 15:48
# author: YANG, HENG <hy345@exeter.ac.uk> (杨恒)
# github: https://github.com/yangheng95
# GScholar: https://scholar.google.com/citations?user=NPq5a_0AAAAJ&hl=en
# ResearchGate: https://www.researchgate.net/profile/Heng-Yang-17/research
# Copyright (C) 2022. All Rights Reserved.


import torch.nn as nn

from pyabsa.networks.bert_mean_pooler import BERTMeanPooler




[docs]
class BERT_MLP(nn.Module):


[docs]
    inputs = ["text_indices"]



    def __init__(self, bert, config):
        super(BERT_MLP, self).__init__()
        self.config = config
        self.bert = bert
        # self.pooler = BertPooler(bert.config)
        # self.pooler = nn.AdaptiveAvgPool1d(1)
        self.pooler = BERTMeanPooler()
        self.dense = nn.Linear(self.config.hidden_dim, self.config.output_dim)
        self.dropout = nn.Dropout(self.config.dropout)
        if self.config.sigmoid_regression:
            self.sigmoid = nn.Sigmoid()
        else:
            self.sigmoid = None



[docs]
    def forward(self, inputs):
        text_raw_indices = inputs[0]
        output = self.bert(text_raw_indices)["last_hidden_state"]
        pooled_out = self.pooler(
            output,
            attention_mask=text_raw_indices.ne(self.config.tokenizer.pad_token_id),
        )

        out = self.dense(pooled_out)
        if self.sigmoid:
            out = self.sigmoid(out)
        return out