Typo delta_reward / delat_reward

apeterswu / RL4NMT

Reinforcement Learning for Neural Machine Translation

187 stars 48 forks source link

Typo delta_reward / delat_reward #2

Closed JoostvDoorn closed 6 years ago

JoostvDoorn commented 6 years ago

I was looking through the code a bit and found that delta_reward had a typo here:

https://github.com/apeterswu/RL4NMT/blob/ab8d9fdceca68c9218b6c3ee3c9bd812c5f9b47a/tensor2tensor/utils/bleu_hook.py#L273 and https://github.com/apeterswu/RL4NMT/blob/ab8d9fdceca68c9218b6c3ee3c9bd812c5f9b47a/tensor2tensor/utils/bleu_hook.py#L293

Don't think this affects the experiments, but the unused_kwargs is pretty dangerous so worth checking.

apeterswu commented 6 years ago

Thanks for your interest and detailed checking. I have fixed the typo and checked the reference, the experiments are not affected. But this is worth, thanks again.