公开(公告)号：EP4435674A2

公开(公告)日：2024-09-25

申请号：EP24187863.6

申请日：2024-07-10

申请人： Beijing Baidu Netcom Science Technology Co., Ltd.

发明人： SHI, Yixuan ,  LI, Wei ,  LIU, Jiachen ,  XIAO, Xinyan

IPC分类号： G06N3/092 ,  G06N3/006 ,  G06N3/084 ,  G06N3/09

CPC分类号： G06N3/006 ,  G06N3/092 ,  G06N3/084 ,  G06N3/09

摘要： A computer-implemented method for training a Text-to-Image model includes: obtaining a first Text-to-Image model and a pre-trained reward model, wherein the first Text-to-Image model is used to generate a corresponding image based on input text, and the pre-trained reward model is used to score a data pair composed of the input text and the corresponding generated image; and adjusting the parameters of the first Text-to-Image model based on the pre-trained reward model and a reinforcement learning policy to obtain a second Text-to-Image model.

公开(公告)号：EP4459481A1

公开(公告)日：2024-11-06

申请号：EP24183160.1

申请日：2024-06-19

申请人： BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

发明人： LIU, Jiachen ,  XIAO, Xinyan ,  WU, Hua ,  LI, Guohao ,  LI, Wei ,  ZHU, Hong ,  SHE, Qiaoqiao ,  LV, Yajuan

IPC分类号： G06F16/535 ,  G06F16/9032

摘要： A computer-implemented image generating method includes: obtaining current dialogue data; determining a requirement type of the user in the current round of dialogue based on the current dialogue data; in response to the requirement type being an image processing requirement, determining an action sequence for implementing the image processing requirement; executing the action sequence to generate a target image; and generating response data corresponding to the user input data based on the target image.

公开(公告)号：EP4191544A1

公开(公告)日：2023-06-07

申请号：EP22209356.9

申请日：2022-11-24

申请人： Beijing Baidu Netcom Science Technology Co., Ltd.

发明人： LI, Wei ,  XIAO, Xinyan ,  LIU, Jiachen

IPC分类号： G06V10/80 ,  G06V20/70

摘要： A method and an apparatus for recognizing a token, an electronic device and a storage medium relate to the field of artificial intelligence technologies such as deep learning and natural language processing. The method includes: obtaining first modal data and second modal data; determining a first token of the first modal data and a second token of the second modal data; determining an associated token between the first token and the second token; and recognizing a target shared token between the first modal data and the second modal data based on the first token, the second token and the associated token. The fine-grained associated fusion of the first token and the second token is implemented based on the associated token, thereby obtaining more accurate and richer cross-modal tokens, effectively improving the generality and generalization of the tokens and effectively enhancing the effect of recognizing a token.