Update relevance_model/model.py
Browse files- relevance_model/model.py +0 -3
relevance_model/model.py
CHANGED
|
@@ -2,8 +2,6 @@ import torch
|
|
| 2 |
from torch import nn
|
| 3 |
from transformers import AutoModel, PreTrainedModel, AutoConfig
|
| 4 |
|
| 5 |
-
# 你可以直接复用你原来的类，稍微改一下使其兼容 HF 的 save_pretrained 更好
|
| 6 |
-
# 但为了保持和你训练时完全一致，最简单的就是保留你原本的写法
|
| 7 |
class BERTDiseaseClassifier(nn.Module):
|
| 8 |
def __init__(self, model_type, num_symps) -> None:
|
| 9 |
super().__init__()
|
|
@@ -16,7 +14,6 @@ class BERTDiseaseClassifier(nn.Module):
|
|
| 16 |
|
| 17 |
def forward(self, input_ids=None, attention_mask=None, token_type_ids=None, **kwargs):
|
| 18 |
outputs = self.encoder(input_ids, attention_mask, token_type_ids)
|
| 19 |
-
# 保持和你训练时完全一致的逻辑
|
| 20 |
x = outputs.last_hidden_state[:, 0, :] # [CLS] pooling
|
| 21 |
x = self.dropout(x)
|
| 22 |
logits = self.clf(x)
|
|
|
|
| 2 |
from torch import nn
|
| 3 |
from transformers import AutoModel, PreTrainedModel, AutoConfig
|
| 4 |
|
|
|
|
|
|
|
| 5 |
class BERTDiseaseClassifier(nn.Module):
|
| 6 |
def __init__(self, model_type, num_symps) -> None:
|
| 7 |
super().__init__()
|
|
|
|
| 14 |
|
| 15 |
def forward(self, input_ids=None, attention_mask=None, token_type_ids=None, **kwargs):
|
| 16 |
outputs = self.encoder(input_ids, attention_mask, token_type_ids)
|
|
|
|
| 17 |
x = outputs.last_hidden_state[:, 0, :] # [CLS] pooling
|
| 18 |
x = self.dropout(x)
|
| 19 |
logits = self.clf(x)
|