-
I try to run some training process using models implemented in Pytorch, which is under `./models_pytorch/classifier_pytorch`. I already successfully go through train and predict process of BERT. Howev…
-
模型现在还是不支持转成pytorch版本嘛，用这份代码在自己的数据领域微调了一版，用脚本转成torch的还是报错
AssertionError: ('Pointer shape torch.Size([128]) and array shape (312,) mismatched', torch.Size([128]), (312,))
-
-
bert 提供了extract_feature.py 文件提取特征表征，请问albert_zh 可以使用该文件来提取特征表征吗？多谢
-
这里logits是不是应该为self.probabilities
-
Can provide links of Albert's model and sentence word model ,Thks.
-
您好，在下游做的一个分类任务， 得到了.pb模型， 利用.pb模型搭建tfserving, ,服务的结果：传入一个单句，进行预测，产看预测时间，预测时间达到了200+ ms 之多， 想问问这个是什么原因
-
您好！
我现在用pytorch_transformers想把您的albert预训练模型转换成pytorch。 官方的bert预训练模型可以转换，但尝试了albert_base_zh与albert_tiny_google_zh都出现相似的出错：
AssertionError Traceback (most recent call last)
…
-
加载albert-base模型时报以下错误：
Shape of variable bert/embeddings/word_embeddings:0 ((21128, 768)) doesn't match with shape of tensor bert/embeddings/word_embeddings ([21128, 128]) from checkpoint reader.
-
RuntimeError: Error(s) in loading state_dict for BertModel:
size mismatch for bert.embeddings.word_embeddings.weight: copying a param with shape torch.Size([21128, 128]) from checkpoint, the shape i…