Recent work demonstrates the potential of multilingual pretraining of creating one model that can be used for various tasks in different languages. Previous work in multilingual pretraining has demonstrated that machine translation systems can be created by finetuning on bitext. In this work, we show that multilingual translation models can be created through multilingual finetuning. Instead of finetuning on one direction, a pretrained model is finetuned on many directions at the same time. Compared to multilingual models trained from scratch, starting from pretrained models incorporates the benefits of large quantities of unlabeled monolingual data, which is particularly important for low resource languages where bitext is not available. We demonstrate that pretrained models can be extended to incorporate additional languages without loss of performance. We double the number of languages in mBART to support multilingual machine translation models of 50 languages. Finally, we create the ML50 benchmark, covering low, mid, and high resource languages, to facilitate reproducible research by standardizing training and evaluation data. On ML50, we demonstrate that multilingual finetuning improves on average 1 BLEU over the strongest baselines (being either multilingual from scratch or bilingual finetuning) while improving 9.3 BLEU on average over bilingual baselines from scratch.

The purpose of this study was to determine the optimum release conditions for the free throw in men's basketball. The study used hundreds of thousands of three-dimensional simulations of basketball trajectories. Five release variables were studied: release height, release speed, launch angle, side angle, and back spin. The free throw shooter was assumed to shoot at 70% and to release the ball 2.134 m (7 ft) above the ground. We found that the shooter should place up to 3 Hz of back spin on the ball, should aim the ball towards the back of the ring, and should launch the ball at 52 degrees to the horizontal. We also found that it is desirable to release the ball as high above the ground as possible, as long as this does not adversely affect the player's launch consistency.

Recent work demonstrates the potential of training one model for multilingual machine translation. In parallel, denoising pretraining using unlabeled monolingual data as a starting point for finetuning bitext machine translation systems has demonstrated strong performance gains. However, little has been explored on the potential to combine denoising pretraining with multilingual machine translation in a single model. In this work, we fill this gap by studying how multilingual translation models can be created through multilingual finetuning.Fintuning multilingual model from a denoising pretrained model incorporates the benefits of large quantities of unlabeled monolingual data, which is particularly important for low resource languages where bitext is rare. Further, we create the ML50 benchmark to facilitate reproducible research by standardizing training and evaluation data. On ML50, we show that multilingual finetuning significantly improves over multilingual models trained from scratch and bilingual finetuning for translation into English. We also find that multilingual finetuning can significantly improve over multilingual models trained from scratch for zero-shot translation on non-English directions. Finally, we discuss that the pretraining and finetuning paradigm alone is not enough to address the challenges of multilingual models for to-Many directions performance.

We present a simple yet effective approach to build multilingual speech-to-text (ST) translation through efficient transfer learning from a pretrained speech encoder and text decoder. Our key finding is that a minimalistic LNA (LayerNorm and Attention) finetuning can achieve zero-shot crosslingual and crossmodality transfer ability by only finetuning 10 ∼ 50% of the pretrained parameters. This effectively leverages large pretrained models at low training cost such as wav2vec 2.0 for acoustic modeling, and mBART for multilingual text generation. This sets a new state-ofthe-art for 36 translation directions (and surpassing cascaded ST for 30 of them) on the large-scale multilingual ST benchmark CoV-oST 2 (Wang et al., 2020b) (+6.4 BLEU on average for En-X directions and +6.7 BLEU for X-En directions). Our approach demonstrates strong zero-shot performance in a many-to-many multilingual model (+5.6 BLEU on average across 28 directions), making it an appealing approach for attaining highquality speech translation with improved parameter and data efficiency.

No abstract