{"id":106018,"date":"2022-12-13T09:36:49","date_gmt":"2022-12-13T07:36:49","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/vypusk-sistemy-mashinnogo-perevoda-opennmt-tf-2-30"},"modified":"2022-12-13T09:36:49","modified_gmt":"2022-12-13T07:36:49","slug":"vypusk-sistemy-mashinnogo-perevoda-opennmt-tf-2-30","status":"publish","type":"post","link":"https:\/\/prohoster.info\/en\/blog\/news\/vypusk-sistemy-mashinnogo-perevoda-opennmt-tf-2-30","title":{"rendered":"Release of the OpenNMT-tf 2.30 translation system","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
The release of the OpenNMT-tf 2.30.0 (Open Neural Machine Translation) translation system has been published, utilizing machine learning techniques. The code for the modules developed by the OpenNMT-tf project is written in Python, uses the TensorFlow library, and is distributed under the MIT license. <\/p>\n
A parallel version of OpenNMT is being developed based on the PyTorch library, which differs in terms of supported features. Moreover, OpenNMT based on PyTorch is presented as more user-friendly and multimodal, while the TensorFlow version is noted for being modular, stable, and capable of leveraging GPU capabilities to accelerate the training process of the neural network. To simplify product distribution, the project is also developing a standalone version of the translator in C++ \u2014 CTranslate2, which uses pretrained models without relying on additional dependencies. <\/p>\n
Models have been prepared for English, German, and Catalan languages; for other languages, a model can be independently created based on the data set from the OPUS project (to train the system, two files are provided \u2014 one with sentences in the source language and the second with a quality translation of these sentences into the target language). <\/p>\n
The project is developed with the involvement of SYSTRAN, a company specializing in machine translation tools, and a group of researchers from Harvard, developing human language models for machine learning systems. The user interface is maximally simplified and requires only specifying the input file with the text and a file for saving the translation result. The extension system allows for additional functionality to be implemented based on OpenNMT, such as auto-referencing, text classification, and subtitle generation. <\/p>\n
In the new version: <\/p>\n
\n
Support for TensorFlow 2.11 has been added, but the new Keras optimizers are not yet supported (the use of tf.keras.optimizers.legacy mode is required).\n
Support for a new branch of the CTranslate2 3.x engine has been added, designed for the efficient execution of models with a transformer architecture.\n
A training parameter for the models, pad_to_bucket_boundary has been added to include additional padding, aligning the block size to values that are multiples of length_bucket_width.\n
Integrated support for chrf and chrf++ metrics from the SacreBLEU project, comparing machine translation with reference human translation.\n
The ctranslate2_spec model attribute has been removed as it is no longer used in CTranslate2. <\/ul>\n