HuiResearch / cail2019_track2

中国法研杯CAIL2019要素抽取任务第三名方案分享
139 stars 30 forks source link

请问如果想使用roberta的话,有哪些地方的代码需要改呢 #9

Open lpl22-github opened 4 years ago

HuiResearch commented 4 years ago

如果你用的是Roberta或Roberta wwm系列的,这些模型其实还是Bert结构,可以直接加载。

m13021933043 邮箱:m13021933043@163.com

Signature is customized by Netease Mail Master

在2020年08月10日 12:41,lpl22-github 写道:

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

lpl22-github commented 4 years ago

emmmm我用的是roberta-large,本来报的错是

NotFoundError: /home/w301/data/lpl/roberta/bert_config.json; No such file or directory

把模型的文件名改了后是:

NotFoundError: Unsuccessful TensorSliceReader constructor: Failed to find any matching files for /home/w301/data/lpl/roberta/bert_model.ckpt