By integrating Back Translation into Google Translate, researchers were able to train models with more monolingual text data for low-resource languages, which significantly improved the fluency of the model output. Back Translation augments parallel training data with synthetic parallel data. When dealing with limited resource languages, researchers use Back Translation and M4 Modelling techniques. The researchers also changed the web searcher from a dictionary-based model to an embedding-based model, which increased the amount of collected data by an average of 29 percent without reducing precision. To improve web crawling, researchers replaced its previous data collection system with a new data miner with more focus on precision than recall, enabling higher quality training data collection from public networks.

Results show that the final hybrid model can achieve a high level of performance with better stability and shorter delay. They combined the transformer encoder and an RNN decoder implemented in TensorFlow framework Lingvo to replace the original RNN-based GNMT system and train the model. Google researchers used a Hybrid Model Architecture to advance both High- and Low-Resource Language translation. Improvements since the implementation of new techniques over the last year are highlighted in green at the end of the animation. BLEU score for Google Translate models since 2007.

Google translate filipino to english