laituan245 / bert-dialog-state-tracking

A Simple but Effective BERT Model for Dialog State Tracking on Resource-Limited Systems

Hi there 👋

I am a computer science Ph.D. student at the University of Illinois at Urbana-Champaign. My work focuses on deep learning for information extraction and natural language understanding. I completed my undergraduate degree in Computer Science at KAIST. I completed my Masters’s degree in Computer Science at the Purdue University.

bert-dialog-state-tracking's Issues

Data preprocessing.

Hello Friend, I am unable to find the procedure of data pre-processing in paper and code. could you please give some information to know about data pre-processing procedure? Thank you.

Loss doasn't converge with the latest version of huggingface/transformers.

I've been trying to reproduce your code. Running your original code produces the results on the paper.
Changing only these two lines:

tokenizer = BertTokenizer.from_pretrained(bert_model)
bert      = BertForSequenceClassification.from_pretrained(bert_model, num_labels=2)

with these from huggingface/transformers

tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
bert      = BertForSequenceClassification.from_pretrained('bert-base-uncased', num_labels=2)

The Loss doesn't converge and the accuracy is very low.
Do you have any idea why this happens?

AttributeError: 'NoneType' object has no attribute 'to'

Traceback (most recent call last):
File "E:\Project\DST\BERT-Dialog-State-Tracking-master21\main.py", line 120, in
main(opts)
File "E:\Project\DST\BERT-Dialog-State-Tracking-master21\main.py", line 75, in main
model.move_to_device(opts)
File "E:\Project\DST\BERT-Dialog-State-Tracking-master21\models.py", line 78, in move_to_device
self.bert.to(args.device)
AttributeError: 'NoneType' object has no attribute 'to'