diff options
author | Marcin Junczys-Dowmunt <marcinjd@microsoft.com> | 2018-11-26 10:45:32 +0300 |
---|---|---|
committer | Marcin Junczys-Dowmunt <marcinjd@microsoft.com> | 2018-11-26 10:45:32 +0300 |
commit | 29346583cfaca11b18ebcf748c293f03e3a82975 (patch) | |
tree | 3e424f418bc36e53f057152d67f7ced2738ffb8a | |
parent | 626b5bde373844b35f806c82d31e12016d30e1b2 (diff) |
fix typos
-rw-r--r-- | training-basics-sentencepiece/README.md | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/training-basics-sentencepiece/README.md b/training-basics-sentencepiece/README.md index b15d996..9b122d7 100644 --- a/training-basics-sentencepiece/README.md +++ b/training-basics-sentencepiece/README.md @@ -73,8 +73,8 @@ sample from https://github.com/rsennrich/wmt16-scripts. We also add the back-translated data from http://data.statmt.org/rsennrich/wmt16_backtranslations/ as desribed in http://www.aclweb.org/anthology/W16-2323. In our experiments, -we get a single model that is a good deal than the ensemble from - the Edinburgh WMT2016 paper. +we get a single model that is a good deal better than the ensemble from +the Edinburgh WMT2016 system submission paper. Assuming one GPU, to execute the complete example type: @@ -87,7 +87,7 @@ No preprocessing is required as the Marian command will train a SentencePiece vo the raw text. Next the translation model will be trained and after convergence, the dev and test sets are translated and evaluated with sacreBLEU. -To use with a different GPUs than device 0 or more GPUs (here 0 1 2 3) use the command below: +To use with a different GPU than device 0 or more GPUs (here 0 1 2 3) use the command below: ``` ./run-me.sh 0 1 2 3 |