marcusau

Results 5 comments of marcusau

基于特定领域的word-bert预训练模型

good question, 我也正打算發問相同問題.,. 因词汇乃根據不同應用景場而改變的.

for NER downstream task, the data format of train.csv/test.csv should be different , right?

so if i have huge corpus, then i can follow your codes and distillate it to smaller model for my downstream tasks ,right? If my dataset has 120000 news articles...

for NER downstream task, the data format of train.csv/test.csv should be different , right?

language of your dataset: Financial news about Hong Kong + China in "traditional Chinese" Tasks: phrase 1 tasks: POS + NER phrase 2 tasks: Knowledge graph building + relationship parser...

for NER downstream task, the data format of train.csv/test.csv should be different , right?

Thanks a lot.. i will try with this

for production environment without GPU support !!

i am trying now... give me 2 days to test it Reply you soon