一個是返回token，一個是返回其在字典中的id，如下
def bert_(): model_name = 'bert-base-chinese' MODEL_PATH = 'D:/xhzy-work/PURE/models/bert-base-chinese/' # a.通過詞典導入分詞器 tokenizer = BertTokenizer.from_pretrained(model_name) # b. 導入配置文件 model_config = BertConfig.from_pretrained(model_name) # 修改配置 model_config.output_hidden_states = True model_config.output_attentions = True # 通過配置和路徑導入模型 bert_model = BertModel.from_pretrained(MODEL_PATH, config=model_config) #sen_code = tokenizer.encode_plus('我不喜歡這世界', '我只喜歡你') sen_code = tokenizer.encode("自然語") print("sen_code",sen_code) sen_code0=tokenizer.tokenize("自然語") print("sen_code0", sen_code0) # input_ids = torch.tensor(tokenizer.encode("自然語")).unsqueeze(0) # print("input_ids",input_ids) # outputs = bert_model(input_ids) # print("outputs",outputs) # sequence_output = outputs[0] # pooled_output = outputs[1] # print("outputs",outputs) # print("sequence_output",sequence_output.shape) ## 字向量 # print("pooled_output",pooled_output.shape) ## 句向量 # print('tokenizer.cls_token',tokenizer.cls_token) if __name__ == '__main__': bert_()