在tf1.15的环境中运行github上slot-gated代码，出现rnn_cell_impl中没有rnn_cell_impl属性的问题。已解决

student_low

于 2022-12-03 19:51:40 发布

阅读量888

点赞数 2

CC 4.0 BY-SA版权

文章标签： tensorflow python

本文链接：https://blog.youkuaiyun.com/student_low/article/details/128165458

本文解决了在使用TensorFlow1.15环境下出现的AttributeError问题，该错误源于Slot-Gated Model训练过程中的代码不兼容。通过调整导入模块的方式，成功解决了_rn_cell_impl._linear未定义的问题。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

Traceback (most recent call last):
File "G:\python_project\Slot-Gated\SlotGated-SLU\train.py", line 188, in <module>
training_outputs = createModel(input_data, len(in_vocab['vocab']), sequence_length, len(slot_vocab['vocab']), len(intent_vocab['vocab']), layer_size=arg.layer_size)
File "G:\python_project\Slot-Gated\SlotGated-SLU\train.py", line 119, in createModel
y = rnn_cell_impl._linear(slot_inputs, attn_size, True)
AttributeError: module 'tensorflow.python.ops.rnn_cell_impl' has no attribute '_linear'

环境配置：tf1.15+w10+anaconda+python3.7+cuda11.2+cudnn8.1.0

win10下给rtx3090用的tensorflow1.15以及c++ dll - 知乎 (zhihu.com)

项目源码地址：MiuLab/SlotGated-SLU: Slot-Gated Modeling for Joint Slot Filling and Intent Prediction (github.com)

解决方法：将源代码

from tensorflow.python.ops import rnn_cell_impl

替换成

from tensorflow_core.contrib.rnn.python.ops import core_rnn_cell as rnn_cell_impl

可以解决问题