roberta custom head test2

This commit is contained in:
s444501 2023-02-12 19:31:34 +01:00
parent 024ad6aae5
commit 9766a22440
2 changed files with 132 additions and 6 deletions

View File

@ -7,28 +7,154 @@ from transformers import RobertaForSequenceClassification, RobertaModel
from transformers.modeling_outputs import SequenceClassifierOutput
class LeakyHeadCustom(nn.Module):
"""Incorporates Leaky ReLU"""
class RobertaLeakyHead(nn.Module):
"""Head for sentence-level classification tasks."""
def __init__(self, config):
super().__init__()
self.use_hidden_states = config.use_hidden_states
hidden_size = config.hidden_size
if self.use_hidden_states:
hidden_size *= 2
self.dense_1 = nn.Linear(hidden_size, 2 * hidden_size)
self.dense_2 = nn.Linear(2 * hidden_size, hidden_size)
self.dense_2 = nn.Linear(2 * hidden_size, 4 * hidden_size)
self.dense_3 = nn.Linear(4 * hidden_size, 2 * hidden_size)
self.dense_4 = nn.Linear(2 * hidden_size, hidden_size)
classifier_dropout = (
config.classifier_dropout if config.classifier_dropout is not None else config.hidden_dropout_prob
)
self.dropout = nn.Dropout(classifier_dropout)
self.out_proj = nn.Linear(hidden_size, config.num_labels)
def forward(self, features, **kwargs):
x = features[:, 0, :] # take <s> token (equiv. to [CLS])
if 'hidden_states' in kwargs and kwargs['hidden_states'] is not None:
if self.use_hidden_states:
x = torch.cat(
(
features[:, 0, :],
# take <s> token (equiv. to [CLS]) from hidden states from second from the end
kwargs['hidden_states'][-2][:, 0, :]
),
dim=1
)
else:
x = features[:, 0, :] + kwargs['hidden_states'][-2][:, 0, :]
del kwargs['hidden_states']
else:
x = features[:, 0, :] # take <s> token (equiv. to [CLS])
if self.use_hidden_states:
x = torch.cat(
(
features[:, 0, :],
torch.zeros(x.size(), dtype=x.dtype, device=x.device)
),
dim=1
)
x = self.dense_1(x)
x = torch.nn.LeakyReLU(x)
x = self.dropout(x)
x = self.dense_2(x)
x = torch.nn.LeakyReLU(x)
x = self.dropout(x)
x = self.dense_3(x)
x = torch.nn.LeakyReLU(x)
x = self.dropout(x)
x = self.dense_4(x)
x = torch.nn.LeakyReLU(x)
x = self.dropout(x)
x = self.out_proj(x)
return x
class RobertaLeaky(RobertaForSequenceClassification):
_keys_to_ignore_on_load_missing = [r"position_ids"]
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.config = config
self.roberta = RobertaModel(config, add_pooling_layer=False)
self.classifier = RobertaLeakyHead(config)
# Initialize weights and apply final processing
self.post_init()
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.FloatTensor] = None,
token_type_ids: Optional[torch.LongTensor] = None,
position_ids: Optional[torch.LongTensor] = None,
head_mask: Optional[torch.FloatTensor] = None,
inputs_embeds: Optional[torch.FloatTensor] = None,
labels: Optional[torch.LongTensor] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple[torch.Tensor], SequenceClassifierOutput]:
r"""
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
`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.roberta(
input_ids,
attention_mask=attention_mask,
token_type_ids=token_type_ids,
position_ids=position_ids,
head_mask=head_mask,
inputs_embeds=inputs_embeds,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states or self.config.use_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
logits = self.classifier(sequence_output, hidden_states=outputs.hidden_states)
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()
if self.num_labels == 1:
loss = loss_fct(logits.squeeze(), labels.squeeze())
else:
loss = loss_fct(logits, 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,
)
# Simple version #
class RobertaClassificationHeadCustomSimple(nn.Module):

View File

@ -49,11 +49,11 @@ from transformers.utils import check_min_version, send_example_telemetry
from transformers.utils.versions import require_version
from roberta import LeakyHeadCustom, RobertaForSequenceClassificationCustomSimple, RobertaForSequenceClassificationCustom, RobertaForSequenceClassificationCustomAlternative
from roberta import RobertaLeaky, RobertaForSequenceClassificationCustomSimple, RobertaForSequenceClassificationCustom, RobertaForSequenceClassificationCustomAlternative
from gpt2 import GPT2ForSequenceClassificationCustomSimple, GPT2ForSequenceClassificationCustom
MODEL_NAME_TO_CLASS = {
'roberta_leaky': LeakyHeadCustom,
'roberta_leaky': RobertaLeaky,
'roberta_simple': RobertaForSequenceClassificationCustomSimple,
'roberta_hidden': RobertaForSequenceClassificationCustom,
'roberta_hidden_v2': RobertaForSequenceClassificationCustomAlternative,