File size: 5,511 Bytes
c95c7b0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
"""
This implementation is based closely on EncT5: https://arxiv.org/abs/2110.08426
This approach tends to perform better than simply appending a classification head
to T5, especially when reducing the number of parameters in T5-like models.
"""

import torch
import copy
from torch import nn
from transformers import T5PreTrainedModel
from transformers.models.t5.modeling_t5 import T5Stack
from transformers.modeling_outputs import SequenceClassifierOutput
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss


def mean_pooling(inputs, mask):
    token_embeddings = inputs
    input_mask_expanded = mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    sum_embeddings = torch.sum(token_embeddings * input_mask_expanded, 1)
    sum_mask = torch.clamp(input_mask_expanded.sum(1), min=1e-9)
    return sum_embeddings / sum_mask


class MeanPooler(nn.Module):
    """ Calcualte simple average of the inputs """
    def __init__(self, input_size=None):
        super().__init__()

    def forward(self, inputs, mask=None):
        if mask is None:
            pooled_output = inputs.mean(dim=1)
        else:
            pooled_output = mean_pooling(inputs, mask)
        return None, pooled_output
 

class AdaptivePooler(nn.Module):
    """ Calcualte weighted average of the inputs with learnable weights """
    def __init__(self, input_size):
        super().__init__()
        self.input_size = input_size
        self.w = nn.Linear(self.input_size, 1, bias=True)

    def forward(self, inputs, mask=None):
        batch_size, seq_len, emb_dim = inputs.shape
        scores = torch.squeeze(self.w(inputs), dim=-1)
        weights = nn.functional.softmax(scores, dim=-1)
        if mask is not None:
            weights = weights * mask
            weights = weights / weights.sum(dim=-1, keepdims=True)
        outputs = (inputs.permute(2, 0, 1) * weights).sum(-1).T
        return weights, outputs
 

class T5ForSequenceClassification(T5PreTrainedModel):
    def __init__(self, config, pooler='adaptive'):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.config = config

        self.shared = nn.Embedding(config.vocab_size, config.d_model)

        encoder_config = copy.deepcopy(config)
        encoder_config.is_decoder = False
        encoder_config.use_cache = False
        encoder_config.is_encoder_decoder = False

        self.encoder = T5Stack(encoder_config, self.shared)

        pooler_class = AdaptivePooler if pooler == 'adaptive' else MeanPooler
        self.pooler = pooler_class(input_size=config.hidden_size)
        self.dropout = nn.Dropout(config.dropout_rate)
        self.classifier = nn.Linear(config.hidden_size, config.num_labels)

        self.init_weights()

        # Model parallel
        self.model_parallel = False
        self.device_map = None

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        head_mask=None,
        inputs_embeds=None,
        output_attentions=None,
        output_hidden_states=None,
        return_dict=None,
        labels=None,
    ):
        r"""
        labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
            Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ...,
            config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss),
            If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        outputs = self.encoder(
            input_ids=input_ids,
            attention_mask=attention_mask,
            head_mask=head_mask,
            inputs_embeds=inputs_embeds,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        weights, pooled_output = self.pooler(outputs[0], mask=attention_mask)
        pooled_output = self.dropout(pooled_output)
        logits = self.classifier(pooled_output)

        loss = None
        if labels is not None:
            if self.config.problem_type is None:
                if self.num_labels == 1:
                    self.config.problem_type = "regression"
                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
                    self.config.problem_type = "single_label_classification"
                else:
                    self.config.problem_type = "multi_label_classification"

            if self.config.problem_type == "regression":
                loss_fct = MSELoss()
                loss = loss_fct(logits.view(-1, self.num_labels), labels)
            elif self.config.problem_type == "single_label_classification":
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
            elif self.config.problem_type == "multi_label_classification":
                loss_fct = BCEWithLogitsLoss()
                loss = loss_fct(logits, labels)
        if not return_dict:
            output = (logits,) + outputs[2:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutput(
            loss=loss,
            logits=logits,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )