Translatium multilingual services5/16/2023 ![]() ![]() While initializing the encoder is straightforward, the decoder is less so, since it adds cross-attention to the encoder’s self-attention. Our starting point was DeltaLM ( DeltaLM: Encoder-Decoder Pre-training for Language Generation and Translation by Augmenting Pretrained Multilingual Encoders), the latest in the increasingly powerful series of massively multilingual pretrained language models from Microsoft.ĭeltaLM is an encoder-decoder model, but instead of training from scratch, it is initialized from a previously pretrained state-of-the-art encoder-only model, specifically ( TULRv3). In this blog post, let’s take a look under the hood at the winning Microsoft ZCode-DeltaLM model. ( Findings of the WMT 2021 Shared Task on Large-Scale Multilingual Machine Translation, Wenzek et al, WMT 2021).įigure 1: Official Results (BLEU scores) on the Full-Task and the Small-Task1 at the WMT 2021 Large Scale Multilingual Translation shared task The ZCode-DeltaLM approach ![]() The Microsoft ZCode-DeltaLM model won all three tasks by huge margins, including an incredible 10+ point gain over the M2M100 model in the large task evaluated on a massive 10,000 language pairs. The Microsoft Translator ZCode team, working together with Turing team and Microsoft Research Asia, competed in the “Large-scale Multilingual Translation” track, which consisted of a Full Task of translating between all 10,000 directions across 101 languages, and two Small tasks: One focused on 5 central and southern European languages, and one on 5 south-east Asian languages. WMT brings together researchers from across the entire Machine Translation field, both industry and academia, to participate in a series of shared tasks, each defining a benchmark in an important area of machine translation to push the field into new frontiers. The annual Conference on Machine Translation (aka WMT 2021) concluded last week in beautiful Punta Cana, Dominican Republic. More recently, Microsoft announced the largest Megatron-Turing NLG 530B parameters model. Recently, the latest Turing universal language representation model ( T-ULRv5), a Microsoft-created model is once again the state of the art and at the top of the Google XTREME public leaderboard at that time. Last summer, we announced our large scale Multi-Lingual Mixture of Expert model with DeepSpeed that can outperform individual large scale bi-lingual models. We continue to push frontiers with Multilingual models to support various language scenarios across Microsoft. The Microsoft Translator ZCode team is working together with Microsoft Project Turing and Microsoft Research Asia to advance language and multilingual support at the core of this initiative. ![]() □□□ En Italia, por ejemplo, la protagonista se llama "Vaiana", mientras que la película recibió el nombre de "Oceania", puesto que Moana en Italia fue una actriz de cine de adultos de los 90 que murió con 33 años (qué pequeño es el mundo).Microsoft is on a quest for AI at Scale with high ambition to enable the next generation of AI experiences. Muchos países europeos como Francia o Ucrania también adoptaron el nombre de "Vaiana" □□ Otros países también cambiaron el nombre original. En su lugar, se eligió "Vaiana", que significa "Agua de la cueva" en tahitiano. ❓□¿Por qué se le cambió el título en español? En España "Moana" es una marca registrada, por lo que Disney tuvo que cambiar el nombre para evitar un posible conflicto. □□ El título original de la película es "Moana", que significa "Mar profundo" en maorí. □□ La traducción del título de las películas no es una tarea de la que se encargue el traductor de la película como mucha gente piensa, pero este caso me parecía digno de comentar. □❗¿Sabías que la película Vaiana ha tenido distintos nombres en distintos países? ![]()
0 Comments
Leave a Reply. |