Issues
- 0
是否可以训练出1024长度的模型
#32 opened by xxWeiDG - 0
请问一下可以替代CRF,进行信息抽取任务吗,需要修改哪些地方
#31 opened by guanxichouqu - 1
每个batch计算P/R/F求平均,其实不等于整体的P/R/F
#27 opened by wushihu - 1
IndexError: list index out of range
#30 opened by 66246764 - 3
预测结果实体多且长的问题
#23 opened by qiutzh - 0
ValueError,预训练模型问题
#29 opened by 66246764 - 0
标签数量
#28 opened by GitMonkey0 - 11
- 5
IndexError: list index out of range
#22 opened by cqcl1 - 1
模型在其他数据集下计算时,F1效果不佳,从何下手?
#24 opened by Joel4U - 1
- 1
loss的参数顺序需要修正
#25 opened by DarthZhu - 0
应用于英文数据
#26 opened by zhanghanweii - 0
- 0
请问如何对未知标签数据进行预测?
#20 opened by cqcl1 - 3
请问gp 算logits的时候 最后为什么要 开方?
#18 opened by CheungZeeCn - 1
当调大batch_size时,比如512,第一个epoch为的验证指标都是0
#15 opened by JeremySun1224 - 1
您好?换了其他数据集,好像会有文本长度过长的问题,报错如下。望回答
#13 opened by Jia1201 - 1
请教下loss计算时是如何考虑padding部分的呢?
#14 opened by tianylin98 - 0
- 0
这个Global Pointer包大家从那里下载呢
#17 opened by Eyeistar - 0
split的维度问题
#16 opened by Alwin4Zhang - 6
大哥,你本人测试的结果也是这样吗
#8 opened by 1a2cjitenfei - 0
- 1
你好,苏神新发布了一个应用与 GP 的loss,请问能实现吗
#11 opened by fuxuelinwudi - 1
为什么f1,precision,recall是step累加的?
#9 opened by rxdaozhang - 1
- 2
请问【cls】与【sep】分隔符可以不需要吗?是有意为之吗
#2 opened by xhw205 - 2
- 1
train.py中为啥每个epoch都要重新定义optimizer?
#7 opened by alkaideemo - 1
去除padding部分,以及最后计算acc
#3 opened by Tongjilibo