公开(公告)号：US20240370663A1

公开(公告)日：2024-11-07

申请号：US18562946

申请日：2022-05-12

Applicant: BEIJING YOUZHUJU NETWORK TECHNOLOGY CO., LTD.

Inventor： Shanbo CHENG ,  Mingxuan WANG ,  Lei LI

IPC: G06F40/40

Abstract: A translation method, a translation apparatus, a translation device, and a storage medium are provided. The method includes: firstly, determining the second source sentence semantically similar to the first source sentence, then determining the target source word used in both of the first source sentence and the second source sentence, and if translated words for the target source word in the first source sentence and the second source sentence are different, determining the target translated word for the target source word according to the probability of the target source word being translated into the first translated word or the second translated word. Thus, the translation method uses not only the information of a second translated sentence for the second source sentence but also the information of the second source sentence. The translation of the first source sentence is corrected according to the information of similar words.

公开(公告)号：US20240296283A1

公开(公告)日：2024-09-05

申请号：US18259210

申请日：2021-12-10

Applicant: BEIJING YOUZHUJU NETWORK TECHNOLOGY CO., LTD.

Inventor： Yang WEI ,  Ying XIONG ,  Xiaohui WANG ,  Mingxuan WANG ,  Lei LI

IPC: G06F40/279 ,  G06F18/213

CPC classification number: G06F40/279 ,  G06F18/213

Abstract: A quantization method and apparatus for a text feature extraction model, and a device and a storage medium. The method includes: in a training process of a text feature extraction model, determining, according to a target quantization parameter, a quantization interval corresponding to the target quantization parameter, where the quantization interval includes a part of floating-point values of the target quantization parameter; constructing a mapping relationship between floating-point values and fixed-point values of the target quantization parameter based on the quantization interval, where a floating-point value smaller than a left end point of the quantization interval—is mapped to a quantized minimum fixed-point value, and a floating-point values larger than a right end point of the quantization interval is mapped to a quantized maximum fixed-point value; and performing a quantization operation on the target quantization parameter based on the mapping relationship.

公开(公告)号：US20240037349A1

公开(公告)日：2024-02-01

申请号：US18255790

申请日：2021-11-17

Applicant: BEIJING YOUZHUJU NETWORK TECHNOLOGY CO., LTD.

Inventor： Chengqi ZHAO ,  Jianze LIANG ,  Mingxuan WANG ,  Lei LI

IPC: G06F40/58 ,  G06F40/51

CPC classification number: G06F40/58 ,  G06F40/51

Abstract: Provided are a model training method and apparatus, a machine translation method and apparatus, a device, and a storage medium. The model training method includes the steps described below. Through a neural network pruning technique, a respective influence degree of each parameter in multiple parameters in a first translation model on a translation result in a first field is determined to obtain at least one first parameter and at least one second parameter. By using the first corpus of the first field, the at least one first parameter is trained obtain the second translation model, and the at least one second parameter remains unchanged. Similarity between a translation result of the second translation model in the first field and a translation result of the first translation model in the first field meets a preset condition.

公开(公告)号：US20240347044A1

公开(公告)日：2024-10-17

申请号：US18752273

申请日：2024-06-24

Applicant: Beijing Youzhuju Network Technology Co., Ltd.

Inventor： Zhicheng LIU ,  Jian TONG ,  Chengqi ZHAO ,  Tao WANG ,  Mingxuan WANG

IPC: G10L15/06 ,  G06F40/58 ,  G10L15/18

CPC classification number: G10L15/063 ,  G06F40/58 ,  G10L15/1815

Abstract: A training method for a speech translation model, a speech translation method, an apparatus, and a device are provided. The training method includes: after entering a model training phase, controlling the speech translation model to execute a speech translation training task; controlling the speech translation model to simultaneously execute an auxiliary training task of the speech translation training task; and adjusting network parameters of the speech translation model according to the speech translation training task and the auxiliary training task, so as to obtain an updated speech translation model after training.

公开(公告)号：US20240265214A1

公开(公告)日：2024-08-08

申请号：US18565946

申请日：2022-07-26

Applicant: Beijing Youzhuju Network Technology Co., Ltd.

Inventor： Zewei SUN ,  Mingxuan WANG ,  Lei LI

IPC: G06F40/58 ,  G06N3/08

CPC classification number: G06F40/58 ,  G06N3/08

Abstract: Embodiments of the present disclosure relate to a translation processing method and apparatus, a device and a medium. The method comprises: generating a multilingual representation model by training according to a monolingual corpus of each language among a plurality of languages, and generating a multilingual generation model according to the monolingual corpus of each language; concatenating the multilingual representation model and the multilingual generation model with a first translation model respectively to generate a target model to be trained; and generating a second translation model by training the target model according to a bilingual corpus among the plurality of languages, and performing translation processing on target information to be processed, according to the second translation model.