This paper describes the Global Tone Communication Co., Ltd.'s submission of the WMT18 shared news translation task. We participated in the English-to-Chinese direction and get the best BLEU (43.8) scores among all the participants. The submitted system focus on data clearing and techniques to build a competitive model for this task. Unlike other participants, the submitted system are mainly relied on the data filtering to obtain the best BLEU score. We do data filtering not only for provided sentences but also for the back translated sentences. The techniques we apply for data filtering include filtering by rules, language models and translation models. We also conduct several experiments to validate the effectiveness of training techniques. According to our experiments, the Annealing Adam optimizing function and ensemble decoding are the most effective techniques for the model training.
This paper describes the Global Tone Communication Co., Ltd.'s submission of the WMT19 shared news translation task. We participate in six directions: English to (Gujarati, Lithuanian and Finnish) and (Gujarati, Lithuanian and Finnish) to English. Further, we get the best BLEU scores in the directions of English to Gujarati and Lithuanian to English (28.2 and 36.3 respectively) among all the participants. The submitted systems mainly focus on backtranslation, knowledge distillation and reranking to build a competitive model for this task. Also, we apply language model to filter monolingual data, back-translated data and parallel data. The techniques we apply for data filtering include filtering by rules, language models. Besides, We conduct several experiments to validate different knowledge distillation techniques and right-to-left (R2L) reranking.
Flow characteristics of stirred tanks with different structures were calculated by taking RNG k-ε model as the turbulent flow model. The results showed that at the same rotational speed, a large number of axial and radial vortexes were formed in the stirred tank with the baffle. The velocity in the blade area was high, and it decreased rapidly with the increasing distance to the blade. The double peak area of the radial velocity was formed in the stirred tank with baffle, and the high and low speed cycles were obtained in the cross-section. The baffle increased not only the axial circulation of the liquid in the tank but also the radial circulation, which help to mix the liquid.
The structures of nozzle outlet have great effects on jet noise generated by submerged gas jet. Five nozzles with different cross-sections were designed to study the influences of nozzle structures on sound levels by experimental methods. The flow characteristics and sound pressure levels were measured and the gas-liquid flows formed by submerged gas jets under different Weber numbers were also observed. The results show that the gas-liquid flows transformed from bubbling regime to jetting regime with the increase of Weber numbers. With respect to bubbling flow, no significant differences in sound levels for different nozzles were found. However, a remarkable increase of acoustic emission was caused by the transformation of flow patterns over the frequency bands of 0.6-4 kHz for all nozzles. In addition, sound pressure levels for elliptical nozzle were much lower than those of other nozzles over the frequency bands of 0.6-4 kHz. Therefore, it would be a reasonable strategy to adopt several elliptical orifices instead of round nozzle to reduce the gas jet noise in the design of underwater exhaust muffler.
In order to study the effect of oil boom gate under different towing speeds, the standard k-ɛ model is combined with the VOF multiphase flow model to analyze the flow field morphology of towage state. The results show that both the monomer oil boom and the double oil boom have good effects in the lower drag speed. But the monomer oil boom would lose its effect gradually with the increasing of drag speed. When the drag speed is up to 0.7 m/s, the monomer oil boom is invalid, while the double oil boom still works well at this speed. It reveals that the double oil boom has better effects than that of monomer oil boom.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.