search

ictnlp / OR-NMT

Source Code for ACL2019 paper <Bridging the Gap between Training and Inference for Neural Machine Translation>
42 stars 10 forks source link

training parameters #6

Closed haorannlp closed 4 years ago

haorannlp commented 4 years ago

Hi Zhang Wen,

Would you mind sharing the detailed training parameters that can reproduce the results for original transformer,oracle transformer in word level oracle transformer in sentence levelin your paper? It took me more than 10days to train a single base model. It is really hard for me to finetune the parameters on this huge dataset. Any help would be greatly appreciated.

zhang-wen commented 4 years ago

@haorannlp Hi, you could finetune based on the pretrained Transformer model for fast training. We will update the parameters.

haorannlp commented 4 years ago

@zhang-wen , could you share the way you implement scheduled sampling in Oracle transformer?

zhang-wen commented 4 years ago

@haorannlp please refer to our code for the details of the implementation.