File size: 3,291 Bytes
fced272 | 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 | # models/text_encoder.py
"""
文本编码器模块：使用PubMedBERT对药物治疗文本进行编码
"""
import torch
import torch.nn as nn
from transformers import AutoTokenizer, AutoModel
class DrugTextEncoder(nn.Module):
"""
使用PubMedBERT编码药物治疗文本
支持冻结/微调模式
"""
def __init__(
self,
model_name: str = "microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext",
freeze: bool = False,
output_dim: int = 768,
max_length: int = 512,
use_pooler_output: bool = True,
):
super().__init__()
self.max_length = max_length
self.use_pooler_output = use_pooler_output
# 加载预训练的tokenizer和模型
self.tokenizer = AutoTokenizer.from_pretrained(model_name)
self.bert = AutoModel.from_pretrained(model_name)
# 是否冻结BERT参数
if freeze:
for param in self.bert.parameters():
param.requires_grad = False
# 投影层（可选，如果需要改变输出维度）
self.projection = None
if output_dim != 768:
self.projection = nn.Linear(768, output_dim)
self.output_dim = output_dim
def forward(self, texts):
"""
Args:
texts: List[str] or str, 药物治疗的JSON字符串
Returns:
embeddings: [B, output_dim]
"""
# 如果是单个字符串，转为列表
if isinstance(texts, str):
texts = [texts]
# Tokenization
encoded = self.tokenizer(
texts,
padding=True,
truncation=True,
max_length=self.max_length,
return_tensors='pt'
)
# 移到正确的设备
device = next(self.bert.parameters()).device
encoded = {k: v.to(device) for k, v in encoded.items()}
# BERT编码
outputs = self.bert(**encoded)
# 旧版 checkpoint 更可能依赖 BERT pooler_output，而不是直接取 CLS hidden state
if self.use_pooler_output and getattr(outputs, "pooler_output", None) is not None:
cls_embeddings = outputs.pooler_output
else:
cls_embeddings = outputs.last_hidden_state[:, 0, :] # [B, 768]
# 可选的投影
if self.projection is not None:
cls_embeddings = self.projection(cls_embeddings)
return cls_embeddings
def unfreeze(self):
"""解冻BERT参数，用于微调"""
for param in self.bert.parameters():
param.requires_grad = True
def freeze(self):
"""冻结BERT参数"""
for param in self.bert.parameters():
param.requires_grad = False
# 测试代码
if __name__ == "__main__":
encoder = DrugTextEncoder(freeze=True)
# 模拟输入
sample_text = '{"pre":{"tp":"T1","therapies":[{"type":"chemotherapy","agents":[{"name":"Temozolomide","dose":"75mg/m2"}]}]}}'
embedding = encoder([sample_text, sample_text])
print(f"Output shape: {embedding.shape}") # [2, 768]
|