如何在BertForSequenceClassification之上添加额外的致密层？

import torch
import torch.nn as nn
from transformers import BertModel

class BertWithDense(nn.Module):
    def __init__(self, num_labels):
        super(BertWithDense, self).__init__()
        self.bert = BertModel.from_pretrained('bert-base-uncased')
        self.dropout = nn.Dropout(0.1)
        self.dense = nn.Linear(self.bert.config.hidden_size, 256)  # 添加一个256维的致密层
        self.relu = nn.ReLU()
        self.classifier = nn.Linear(256, num_labels)  # 根据任务的类别数定义分类器层

    def forward(self, input_ids, attention_mask):
        outputs = self.bert(input_ids=input_ids, attention_mask=attention_mask)
        pooled_output = outputs.pooler_output
        pooled_output = self.dropout(pooled_output)
        dense_output = self.dense(pooled_output)
        dense_output = self.relu(dense_output)
        logits = self.classifier(dense_output)

        return logits

model = BertWithDense(num_labels=2)  # 假设有2个类别需要分类