what is the hparams_set for benchmark transformer model?

THUNLP-MT / THUMT

An open-source neural machine translation toolkit developed by Tsinghua Natural Language Processing Group

BSD 3-Clause "New" or "Revised" License

701 stars 197 forks source link

what is the hparams_set for benchmark transformer model? #100

Closed stevensgeek41 closed 3 years ago

stevensgeek41 commented 3 years ago

Hi, I want to reproduce the benchmark result for Zh-En task, may I ask what is the hparams_set parameter for the transformer model? I see there are two options, "big" and "big_v2", which one did you use? Thanks!

stevensgeek41 commented 3 years ago

I was able to get BLEU score of 22.06 with base transformer model following walkthrough markdown. It's pretty solid.