nlp-notebook
nlp-notebook copied to clipboard
jasoncao11
3-2 'tuple' object has no attribute 'last_hidden_state'
(pytorch18) z@z:~/code/nlp-notebook-master/3-2.Bert-CRF$ python demo_train.py
Some weights of the model checkpoint at ./bert-base-chinese were not used when initializing BertForNER: ['cls.predictions.bias', 'cls.predictions.transform.dense.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.decoder.weight', 'cls.seq_relationship.weight', 'cls.seq_relationship.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.LayerNorm.bias']
This IS expected if you are initializing BertForNER from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPretraining model).
This IS NOT expected if you are initializing BertForNER from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).
Some weights of BertForNER were not initialized from the model checkpoint at ./bert-base-chinese and are newly initialized: ['transitions', 'hidden2label.weight', 'hidden2label.bias']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
[Train Epoch 0]: 0%| | 0/1584 [00:00<?, ?it/s]
Traceback (most recent call last):
File "demo_train.py", line 66, in
输出如上，尝试修改model.from_pretrained（model_path，output_hidden_states = True）也不行 请问是哪里出了问题？环境配置是一样的
估计是BERT的版本不匹配 我刚刚也遇到了这个问题，试了下sequence_output = x[0]就可以了。 猜测原因如下： bert的输出包括last_hidden_state，pooler_output，hidden_states，attentions，默认只输出前两项，想要输出后两项需要像你说的那样修改output_hidden_states = True和output_attentions = True 代码中需要取的是last_hidden_state即最后一层的隐状态，而该版本似乎将其作为x的一个属性。而我的版本将x处理成了一个元组，故取了x的第一个元素。
估计是BERT的版本不匹配 我刚刚也遇到了这个问题，试了下sequence_output = x[0]就可以了。 猜测原因如下： bert的输出包括last_hidden_state，pooler_output，hidden_states，attentions，默认只输出前两项，想要输出后两项需要像你说的那样修改output_hidden_states = True和output_attentions = True 代码中需要取的是last_hidden_state即最后一层的隐状态，而该版本似乎将其作为x的一个属性。而我的版本将x处理成了一个元组，故取了x的第一个元素。
谢谢！我目前可以成功训练出模型，但是不知道为什么运行demo_eval.py预测的实体全部为空……