siat-nlp / mams-for-absa Goto Github PK

Hi,

According to the paper, there is a large train dataset for the ACSA of ~7K sentences. What I found here is the smaller version of ~3K instances. Can you share the large version, please?

Regards,
Omnia

Hi, I,m interested in your paper, where can I read your paper?

has anyone find the log.yml?
i couldnt find it : ( thus i couldnt run the code...

Excuse me, I met this problem running train.py.
Could you plz tell me if I did anything wrong?
I really appreciate your reply :)

File "../aspect_category_model/bert_capsnet.py", line 65, in forward
sentence_mask = segment_mask & (1 - bert_segment).byte()
RuntimeError: result type Byte can't be cast to the desired output type Bool

Hello, I would like to implement it in Korean.
If I label the data set created in Korean, can I use it as it is?

Dear authors,

I really like your work. And I feel sad that no work has started citing your dataset and paper. In order to make your MAMS dataset of higher impact, I am doing a follow-up paper. However, outputs of all models in your table 3 will be an important component in our Analysis section. Do you by chance still keep these outputs of all models of MAMS and Restaurant?

These data will significantly speed up the work in our paper, and enlarge the impact of your work. (And even partial outputs will help.) Thank you!

Hi,

I tried to run the bert-capsnet model on my own data using bert-base-multilingual-uncased but I got this error

Could someone help plz?

Thank u

Excuse me, how did you get this file sentiment.json?

In the ./train/make_aspect_category_model.py, the function make_recurrent_capsule_network() randomly initialized a variable aspect_embedding，but never use it in the network.

Is this a bug？

你好，我有点不理解数据中的from和to代表着什么呢？希望可以得到解答。谢谢

例如：
When I got home, there was a message on the machine because the owner realized that our waitress forgot to charge us for our wine.

Hello, how could we use a trained model for inference on an unlabeled dataset?

Dear authors,

Great work, it could be very nice to provide the code for the other baseline models in your paper, e.g. AOA LSTM, GCAE. This will help people reproduce the result in table 3.

thanks