Delete bert_embeddings.py
Browse files- bert_embeddings.py +0 -82
bert_embeddings.py
DELETED
@@ -1,82 +0,0 @@
|
|
1 |
-
import logging
|
2 |
-
from typing import Optional
|
3 |
-
|
4 |
-
import torch
|
5 |
-
from torch.nn.modules.utils import consume_prefix_in_state_dict_if_present
|
6 |
-
from transformers import BertPreTrainedModel
|
7 |
-
|
8 |
-
from bert_layers_mosa import BertModel
|
9 |
-
|
10 |
-
logger = logging.getLogger(__name__)
|
11 |
-
|
12 |
-
|
13 |
-
class MosaicBertForEmbeddingGeneration(BertPreTrainedModel):
|
14 |
-
|
15 |
-
def __init__(self, config, add_pooling_layer=False):
|
16 |
-
"""
|
17 |
-
Initializes the BertEmbeddings class.
|
18 |
-
|
19 |
-
Args:
|
20 |
-
config (BertConfig): The configuration for the BERT model.
|
21 |
-
add_pooling_layer (bool, optional): Whether to add a pooling layer. Defaults to False.
|
22 |
-
"""
|
23 |
-
super().__init__(config)
|
24 |
-
assert (
|
25 |
-
config.num_hidden_layers >= config.num_embedding_layers
|
26 |
-
), "num_hidden_layers should be greater than or equal to num_embedding_layers"
|
27 |
-
self.config = config
|
28 |
-
self.strategy = config.strategy
|
29 |
-
self.bert = BertModel(config, add_pooling_layer=add_pooling_layer)
|
30 |
-
# this resets the weights
|
31 |
-
self.post_init()
|
32 |
-
|
33 |
-
@classmethod
|
34 |
-
def from_pretrained(
|
35 |
-
cls, pretrained_checkpoint, state_dict=None, config=None, *inputs, **kwargs
|
36 |
-
):
|
37 |
-
"""Load from pre-trained."""
|
38 |
-
# this gets a fresh init model
|
39 |
-
model = cls(config, *inputs, **kwargs)
|
40 |
-
|
41 |
-
# thus we need to load the state_dict
|
42 |
-
state_dict = torch.load(pretrained_checkpoint)
|
43 |
-
# remove `model` prefix to avoid error
|
44 |
-
consume_prefix_in_state_dict_if_present(state_dict, prefix="model.")
|
45 |
-
missing_keys, unexpected_keys = model.load_state_dict(state_dict, strict=False)
|
46 |
-
|
47 |
-
if len(missing_keys) > 0:
|
48 |
-
logger.warning(
|
49 |
-
f"Found these missing keys in the checkpoint: {', '.join(missing_keys)}"
|
50 |
-
)
|
51 |
-
|
52 |
-
logger.warning(f"the number of which is equal to {len(missing_keys)}")
|
53 |
-
|
54 |
-
if len(unexpected_keys) > 0:
|
55 |
-
logger.warning(
|
56 |
-
f"Found these unexpected keys in the checkpoint: {', '.join(unexpected_keys)}",
|
57 |
-
)
|
58 |
-
logger.warning(f"the number of which is equal to {len(unexpected_keys)}")
|
59 |
-
|
60 |
-
return model
|
61 |
-
|
62 |
-
def forward(
|
63 |
-
self,
|
64 |
-
input_ids: Optional[torch.Tensor] = None,
|
65 |
-
attention_mask: Optional[torch.Tensor] = None,
|
66 |
-
token_type_ids: Optional[torch.Tensor] = None,
|
67 |
-
position_ids: Optional[torch.Tensor] = None,
|
68 |
-
subset_mask: Optional[torch.Tensor] = None,
|
69 |
-
output_all_encoded_layers: Book = True,
|
70 |
-
) -> torch.Tensor:
|
71 |
-
|
72 |
-
embedding_output = self.bert.embeddings(input_ids, token_type_ids, position_ids)
|
73 |
-
|
74 |
-
encoder_outputs_all = self.bert.encoder(
|
75 |
-
embedding_output,
|
76 |
-
attention_mask,
|
77 |
-
output_all_encoded_layers=output_all_encoded_layers,
|
78 |
-
subset_mask=subset_mask,
|
79 |
-
)
|
80 |
-
|
81 |
-
# batch_size, hidden_dim
|
82 |
-
return encoder_outputs_all
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|