search

ruidan / IMN-E2E-ABSA

Code and dataset for ACL2019 ‘‘An Interactive Multi-Task Learning Network for End-to-End Aspect-Based Sentiment Analysis’’.
Apache License 2.0
99 stars 18 forks source link

TypeError: Layer sequential_1_input does not support masking, but was passed an input_mask: Tensor("concatenate_1/All:0", shape=(?, 78), dtype=bool) #4

Open ConnieTong opened 4 years ago

ConnieTong commented 4 years ago

Hi, I am interested in your work. I run your code, but I encountered the error which is "TypeError: Layer sequential_1_input does not support masking, but was passed an input_mask: Tensor("concatenate_1/All:0", shape=(?, 78), dtype=bool) " when building AE model. My tensorflow version is 1.10.0 Anyone can help me ? Thanks so much ! Bests, Connie Tong

ConnieTong commented 4 years ago

In model.py, line 230, I met the problem.
if args.aspect_layers > 0: aspect_output = aspect_cnn(aspect_output)

ruidan commented 4 years ago

You can check the dependency section in README for package versions.

ConnieTong commented 4 years ago

Thanks. Fixed it.

dongyang12138 commented 4 years ago

Thanks. Fixed it.

请问您最终使用的tensorflow版本是什么，我改成python3代码后没有找到tensorflow1.4.1版本 What is the version of tensorflow you finally used, I did not find the version of tensorflow 1.4.1 after I changed the code to python3

ruidan commented 4 years ago

@dongyang12138 目前代码只在python2的环境测试过， 直接换成python3可能有些问题。

dongyang12138 commented 4 years ago

您好，感谢您的回复，我现在遇到一个问题就是训练集loss全为0，验证集缺正常，然后在后面就出现了除0错误，请问您知道是哪里出了问题吗？ 感激来信。 2020-07-31 17:36:33,927 INFO [Train] loss: 0.0000, [Sentiment] loss: 0.0000, [Domain] loss: 0.0000, [Sentiment] acc: 0.0000, [Domain] acc: 0.0000, 2020-07-31 17:38:20,377 INFO [Validation] loss: 0.5878, [Sentiment] loss: 0.5649, [Domain] loss: 0.0229, [Sentiment] acc: 0.7527, [Domain] acc: 0.9927,

在 2020-07-30 12:11:04，"Ruidan He" notifications@github.com 写道：

@dongyang12138 目前代码只在python2的环境测试过， 直接换成python3可能有些问题。

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

dongyang12138 commented 4 years ago

谢谢，我知道问题出在哪里了，我把python2代码转python3的时候，除号“/”转换成了“//”，成了取整除，所以结果为0了 还有个问题想问您，您的代码中做的除法是不是都是保留小数的除法？

在 2020-07-31 19:26:43，"董洋" dongyang12138@163.com 写道：

您好，感谢您的回复，我现在遇到一个问题就是训练集loss全为0，验证集缺正常，然后在后面就出现了除0错误，请问您知道是哪里出了问题吗？ 感激来信。 2020-07-31 17:36:33,927 INFO [Train] loss: 0.0000, [Sentiment] loss: 0.0000, [Domain] loss: 0.0000, [Sentiment] acc: 0.0000, [Domain] acc: 0.0000, 2020-07-31 17:38:20,377 INFO [Validation] loss: 0.5878, [Sentiment] loss: 0.5649, [Domain] loss: 0.0229, [Sentiment] acc: 0.7527, [Domain] acc: 0.9927,

在 2020-07-30 12:11:04，"Ruidan He" notifications@github.com 写道：

@dongyang12138 目前代码只在python2的环境测试过， 直接换成python3可能有些问题。

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.