413542484 comments

Results 7 comments of

413542484

visual_feedback

Have you solved it? I have the same problem

Using layoutlmv3 to carry out NER task, in the result of model predict, some entities only have 'I-' and no 'B-'

> @413542484 我也遇到这个问题，你解决了吗没解决，我在发现其他ner模型存在同样的问题，应该就是普通的模型预测错误。可以写一个简单的策略处理这种情况。

Using layoutlmv3 to carry out NER task, in the result of model predict, some entities only have 'I-' and no 'B-'

> The explanation here is very clear，The '-- task' parameter is the NER task by default and it can be left blank.

Using layoutlmv3 to carry out NER task, in the result of model predict, some entities only have 'I-' and no 'B-'

> 如何使用 layoutlmv3 训练 NER 任务？你能分享参考链接吗？ https://github.com/microsoft/unilm/blob/master/layoutlmv3/README.md The explanation here is very clear，The '-- task' parameter is the NER task by default and it can be left blank

Using layoutlmv3 to carry out NER task, in the result of model predict, some entities only have 'I-' and no 'B-'

> 可以详细解释一下么，我确实改动了他dataload的代码，不过我在二分类问题上没出现这种情况，在多分类的时候才出现这个问题，你是dataload哪一部分出了问题

model link is wrong

Can you share the trained model again?

您好，模型在训练过程中，验证结果为0

> 参考#5 你好，我在调试代码时发现，tbrec_model.py中test函数里cls_pred_score的值会经常出现为nan的情况，这nan值最终可以追溯到参数pdl、pdt为0并做了分母，这种情况是正常的么？程序并没有报错。