microsoft/MPNet issues and pull requests

#20 - Access to the released fairseq checkpoint

Issue - State: closed - Opened by GeorgeSanchezTR over 1 year ago - 4 comments

#19 - AssertionError: cannot specify cutoff larger than vocab size

Issue - State: open - Opened by xx-m over 1 year ago

#18 - Inconsistencies between data collator output and masked permute in original paper

Issue - State: open - Opened by ghost over 2 years ago

#17 - rgument --arch/-a: invalid choice: 'mpnet_rel_base'这是为什么？

Issue - State: open - Opened by gf52 over 2 years ago

#16 - How to use deepspeed?

Issue - State: open - Opened by LifeIsStrange almost 3 years ago

#15 - The future is to combine MPNet with other language models innovations

Issue - State: open - Opened by LifeIsStrange almost 3 years ago

#14 - SQUAD2.0 test数据集结果

Issue - State: closed - Opened by JunnYu over 3 years ago

#13 - does it support cuda

Issue - State: closed - Opened by tempbrucefu over 3 years ago - 1 comment

#12 - The exact English pretraining data and Chinese pretraining data that are exact same to the BERT paper's pretraining data.

Issue - State: closed - Opened by guotong1988 over 3 years ago - 1 comment

#11 - How to continue pretraining from the released checkpoint?

Issue - State: open - Opened by saroyehun almost 4 years ago - 3 comments

#10 - Setting 'max_len_single_sentence' is now deprecated. This value is automatically set up.

Issue - State: open - Opened by bhadreshpsavani almost 4 years ago - 1 comment

#9 - Training on SQUAD2 gives less results on Evaluation set (Research Paper shows better results)

Issue - State: closed - Opened by bhadreshpsavani almost 4 years ago - 1 comment

#8 - would you release the code about convert the mpnet to the transformers one?

Issue - State: closed - Opened by RyanHuangNLP almost 4 years ago - 3 comments

#7 - 关于MPNet在GLUE dev集合上结果的疑问？

Issue - State: closed - Opened by PROoshio about 4 years ago - 1 comment

#6 - 可否圈下重点代码部分，fairseq代码太多

Issue - State: closed - Opened by guotong1988 about 4 years ago - 1 comment

#5 - Is there results on GLUE for MPNet which is pre-trained on 16G corpora?

Issue - State: open - Opened by PROoshio about 4 years ago

#4 - MPNet如何训练长序列分类任务IMDB和RACE呢？

Issue - State: closed - Opened by PROoshio over 4 years ago - 1 comment

#3 - 关于SQuAD 2.0的评估

Issue - State: closed - Opened by PROoshio over 4 years ago - 6 comments

#2 - input mode questions?

Issue - State: closed - Opened by RyanHuangNLP over 4 years ago - 1 comment

#1 - could you release the finetune performance with fewer epoch on glue?