Closed LibertFan closed 4 years ago
You can try more data, Here is one of our used data with 50M monolingual data: data. Besides, beam search and length penalty are also important for the final result.
Coulld you provide some details about the hyperparameters, beam search and length penalty, etc.? @StillKeepTry That would be very helpful!
I use all data from newcrawl 07-17 for reproducing unsupMT-EnDe. I get bleus en-de:33.3, de-en:26.5, which are still lower than reported results.
I finetune the mass_ende_1024.pth you provide in Unsupvised-MASS with the following command:
The result is:
test_de-en_mt_bleu is about 33.0, which is far behind the 35.2 in your paper.
Could you provide any suggestions to improve my finetuning?