# Neural Machine Translation

Firefox Translations Models is a set of CPU-optimized neural machine translation models developed by Mozilla, designed for the translation feature of the Firefox browser. The model provides fast and accurate translation services, supporting multiple language pairs, through efficient CPU acceleration technology. Its main advantages include high performance, low latency, and support for multiple languages. This model is the core technology of the Firefox browser's translation function, providing users with a seamless web translation experience.

DRT-o1-14B is a neural machine translation model designed to enhance the depth and accuracy of translations through long-chain reasoning. The model extracts English sentences containing metaphors or idioms and employs a multi-agent framework (comprising translator, advisor, and evaluator) to construct long-thought machine translation samples. Trained on the Qwen2.5-14B-Instruct backbone, this model has 14.8 billion parameters and supports BF16 tensor types. Its significance lies in its ability to handle complex translation tasks, especially where deep understanding and reasoning are required, providing a novel solution.

DRT-o1-7B is dedicated to successfully applying long-form reasoning in neural machine translation (MT). The model identifies English sentences suitable for long-form reasoning translation and proposes a multi-agent framework involving three roles: the translator, advisor, and evaluator, to synthesize MT samples. DRT-o1-7B and DRT-o1-14B are trained using Qwen2.5-7B-Instruct and Qwen2.5-14B-Instruct as backbone networks. The main advantage of this model lies in its capability to handle complex linguistic structures and deep semantic understanding, which is crucial for enhancing the accuracy and naturalness of machine translation.

DRT-o1 is a neural machine translation model that optimizes the translation process using extended reasoning chains. It extracts English sentences containing metaphors or similes and employs a multi-agent framework (including translators, advisors, and evaluators) to collaborate and construct extended reasoning machine translation samples. The DRT-o1-7B and DRT-o1-14B models are large language models trained based on Qwen2.5-7B-Instruct and Qwen2.5-14B-Instruct. The primary advantage of DRT-o1 lies in its ability to handle complex linguistic structures and deep semantic understanding, which is crucial for improving the accuracy and naturalness of machine translations.

Lingvanex is a multilingual translation tool that can translate text, voice, images, and files. Leveraging neural machine translation technology, it enhances your work efficiency.