search

chiayewken / Span-ASTE

Code Implementation of "Learning Span-Level Interactions for Aspect Sentiment Triplet Extraction".
MIT License
169 stars 45 forks source link

Error running demo #25

Closed Pinerush closed 2 years ago

Pinerush commented 2 years ago

I get an error running the 3rd cell in the demo saying "pred_out.json" does not exist

chiayewken commented 2 years ago

Hi, can you provide more details on your specific hardware setup and environment? Currently, I'm unable to reproduce your error in the demo colab notebook.

Pinerush commented 2 years ago

It was an error on the Jupiter notebook, worked fine on colab

rdyzakya commented 2 years ago

Oh yes, I found this error too if I change the random seed

dinhgiangltk commented 2 years ago

Hi @chiayewken, I have used the Google Colab notebook to run the demo codes. And the mentioned-above error has appeared. I guested that because the JSON file was opened without a defined mode, then I tried to edit it to be mode 'w'. But the same error was still there. Could you please help me with that? image

dinhgiangltk commented 2 years ago

Hi @rdyzakya, you mean if you do not change the random seed (random_seed=0), you do not get the error, do you?

rdyzakya commented 2 years ago

@dinhgiangltk if I recall, yes as you said, but I try it just now and it was okay. This become a problem when I use another dataset

dinhgiangltk commented 2 years ago

@rdyzakya but I ran the same codes and the error appeared. :(

dipanmoy commented 2 years ago

Yes, until this morning everything was fine. But now it's giving this pred_out file is missing error. Also, unable to retrain model using our own dataset. The training runs for sometime and then stops just before the epoch phase starts. Also doesn't throw up any error.

chiayewken commented 2 years ago

Hi everyone, sorry for the delay. I tried to run the colab demo and change the random seed for training (eg from 0 to 4) but it runs smoothly and there is no error. If your training stage doesn't work, please save the entire terminal log and attach the log file here as the allennlp error may be quite far up.

For those who face the error in cell 3 for the pretrained weights, can you check that the pretrained_14lap weights file is downloaded? If you still face the error, please delete the local files and try again. If possible, it will be useful to know your specific hardware setup and environment.

dipanmoy commented 2 years ago

@chiayewken I was able to run all the cells as well as train my own data yesterday.