Howardqlz

Hi, i am wonder about this loss: ` class_loss = class_criterion(class_logit, target_var) / minibatch_size ` since this loss ignore some samples(no_lable), why here still use the minibatch_size not the labeled_size?

About the vocabulary constructed on snli

2

I see this code: `self.TEXT.build_vocab(self.train, self.dev, self.test, vectors=GloVe(...))` As i know, we should constrcuct vocabulary only on trainset?

Model I am using LayoutLMv3 on Article Regions dataset

1

**Describe** Model I am using LayoutLmv3: I downloaded base model, and use the config file based on Publaynet. I training the model on [Article Regions dataset](https://github.com/cxsoto/article-regions),a little object detection dataset...

About the training step

1

Hello, i want to know how to training my own dataset. Becasue the training_code contains so many files, i dont know the detail steps of training. Can you give a...

code about ESIM

1

Is this problem about the torchtext?? ImportError: cannot import name 'load_word_vectors'

This lead to a very confused result during training and it's difficult to locate this hidden problem...Need modify the categories order to keep them same. Though this is a small...

First thanks for your contribution of this implement of DCN+ by pytorch, and it's really useful for me to understand this task. However, during running the model, i met some...