公开(公告)号：US20250166243A1

公开(公告)日：2025-05-22

申请号：US18513071

申请日：2023-11-17

Applicant: ADOBE INC.

Inventor： Aishwarya Agarwal ,  Srikrishna Karanam ,  Tripti Shukla ,  Balaji Vasan Srinivasan

IPC: G06T11/00 ,  G06F40/284 ,  G06N3/0455 ,  G06N3/084 ,  G06T11/60

Abstract: An image generation model obtains a text prompt, a first attribute token, and a second attribute token. A first set of layers of the image generation model and a first set of time-steps are identified for the first attribute token and a second set of layers of the image generation model and a second set of time-steps are identified for the second attribute token. A synthetic image is generated based on the text prompt, the first attribute token, and the second attribute token by providing the first attribute token to the first set layers of the image generation model during the first set of time-steps and providing the second attribute token to the second set of layers of the image generation model during the second set of time-steps.

公开(公告)号：US12198048B2

公开(公告)日：2025-01-14

申请号：US17153130

申请日：2021-01-20

Applicant: Adobe Inc.

Inventor： Hrituraj Singh ,  Jatin Lamba ,  Denil Pareshbhai Mehta ,  Balaji Vasan Srinivasan ,  Anshul Nasery ,  Aishwarya Agarwal

IPC: G06F16/24 ,  G06F16/242 ,  G06F18/214 ,  G06F18/22 ,  G06F40/20 ,  G06N3/045 ,  G06N3/08 ,  G06V30/40

Abstract: In some embodiments, a multimodal computing system receives a query and identifies, from source documents, text passages and images that are relevant to the query. The multimodal computing system accesses a multimodal question-answering model that includes a textual stream of language models and a visual stream of language models. Each of the textual stream and the visual stream contains a set of transformer-based models and each transformer-based model includes a cross-attention layer using data generated by both the textual stream and visual stream of language models as an input. The multimodal computing system identifies text relevant to the query by applying the textual stream to the text passages and computes, using the visual stream, relevance scores of the images to the query, respectively. The multimodal computing system further generates a response to the query by including the text and/or an image according to the relevance scores.

公开(公告)号：US20240428468A1

公开(公告)日：2024-12-26

申请号：US18337634

申请日：2023-06-20

Applicant: Adobe Inc.

Inventor： Aishwarya Agarwal ,  Srikrishna Karanam ,  Joseph Koonthanam Jose ,  Apoorv Umang Saxena ,  Koustava Goswami ,  Balaji Vasan Srinivasan

IPC: G06T11/00 ,  G06N3/0455

Abstract: The present disclosure relates to systems, methods, and non-transitory computer-readable media that utilizes attention segregation loss and/or attention retention loss at inference time of a diffusion neural network to generate a text-conditioned image. In particular, in some embodiments, the disclosed systems utilize the attention segregation loss to reduce overlap between concepts by comparing attention maps for multiple concepts of a text query corresponding to a denoising step. Further, in some embodiments, the disclosed systems utilize the attention retention loss to improve information retention for concepts across denoising steps by comparing attention maps between different denoising steps. Accordingly, in some embodiments, by utilizing the attention segregation loss and the attention retention loss, the disclosed systems accurately maintain multiple concepts from a text query when generating a text-conditioned image.

公开(公告)号：US11954431B2

公开(公告)日：2024-04-09

申请号：US17522790

申请日：2021-11-09

Applicant: Adobe Inc.

Inventor： Suryateja Bv ,  Vishwa Vinay ,  Niyati Himanshu Chhaya ,  Navita Goyal ,  Elaine Chao ,  Balaji Vasan Srinivasan ,  Aparna Garimella

IPC: G06F40/194 ,  G06F16/901 ,  G06F16/93 ,  G06F40/14

CPC classification number: G06F40/194 ,  G06F16/9027 ,  G06F16/93 ,  G06F40/14

Abstract: Embodiments are disclosed for generating an intelligent change summary are described. In some embodiments, a method of generating an intelligent change summary includes obtaining a representation of a plurality of versions of a document, determining a distance score based on a comparison of a first of version of the document and a second version of the document, the distance score representing a magnitude of changes made from the first version of the document to the second version of the document, and generating a change summary of the document based on the distance score.

公开(公告)号：US11657225B2

公开(公告)日：2023-05-23

申请号：US17348257

申请日：2021-06-15

Applicant: Adobe Inc.

Inventor： Balaji Vasan Srinivasan ,  Kushal Chawla ,  Mithlesh Kumar ,  Hrituraj Singh ,  Arijit Pramanik

IPC: G06F40/284 ,  G06N20/00

CPC classification number: G06F40/284 ,  G06N20/00

Abstract: Systems and methods for generating a tuned summary using a word generation model. An example method includes receiving, at a decoder of the word generation model, a training data learned subspace representation of training data. The method also includes identifying tunable linguistic characteristics of the word generation model and training the decoder to output a training tuned summary of the training data learned subspace representation based on at least one of the tunable linguistic characteristics. The method further includes receiving an input text and a target characteristic token, and generating, by the trained decoder of the word generation model, each word of a tuned summary of the input text from a learned subspace representation and from feedback about preceding words of the tuned summary, wherein the tuned summary is tuned to target characteristics represented by the target characteristic token.

公开(公告)号：US20230121711A1

公开(公告)日：2023-04-20

申请号：US17501602

申请日：2021-10-14

Applicant: Adobe Inc.

Inventor： Niyati Himanshu Chhaya ,  Udit Kalani ,  Roodram Paneri ,  Sreekanth Reddy ,  Niranjan Kumbi ,  Navita Goyal ,  Balaji Vasan Srinivasan ,  Ayush Agarwal

IPC: G06F40/40 ,  G06N3/08 ,  G06F40/279

Abstract: A content generator system receives a request to generate content for a target entity, and one or more keywords. The content generator system retrieves, for the target entity, a current stage identifier linking the target entity to a current stage within a multi-stage objective. The content generator system generates an input vector including the current stage identifier, a target stage identifier, a token embedding comprising the one or more keywords, and a position embedding for each of the one or more keywords, the target stage identifier associated with a target stage within the multi-stage objective different from the current stage. The content generator system generates output text content for the target entity by applying a generative transformer network to the input vector. The content generator system transmits the output text content to a computing device associated with the target entity.

公开(公告)号：US11580307B2

公开(公告)日：2023-02-14

申请号：US16825864

申请日：2020-03-20

Applicant: Adobe Inc.

Inventor： Niyati Himanshu Chhaya ,  Sopan Khosla ,  Balaji Vasan Srinivasan

IPC: G06F16/35 ,  G06F40/30 ,  G06N3/04 ,  G06F40/289 ,  G06N3/08

Abstract: A digital attribution system is described to generate predictions of word attributions from subject data, e.g., titles, subject lines of emails, and so on. To do so, an attribution score is first generated by the digital attribution system that describe an amount to which respective words in the subject data cause performance of a corresponding outcome. The attribution scores are then used by the digital attribution system to generate representations for display in a user interface for respective words in the subject data and may also be used to generate attribution recommendations of changes to be made to the subject data.

公开(公告)号：US11487971B2

公开(公告)日：2022-11-01

申请号：US17073258

申请日：2020-10-16

Applicant: Adobe Inc.

Inventor： Navita Goyal ,  Balaji Vasan Srinivasan ,  Anandha velu Natarajan ,  Abhilasha Sancheti

IPC: G06F17/00 ,  G06K9/62 ,  G06F40/205 ,  G06K9/00 ,  G06V30/414 ,  G06F40/40

Abstract: In some embodiments, a style transfer computing system generates a set of discriminator models corresponding to a set of styles based on a set of training datasets labeled for respective styles. The style transfer computing system further generates a style transfer language model for a target style combination that includes multiple target styles from the set of styles. The style transfer language model includes a cascaded language model and multiple discriminator models selected from the set of discriminator models. The style transfer computing system trains the style transfer language model to minimize a loss function containing a loss term for the cascaded language model and multiple loss terms for the multiple discriminator models. For a source sentence and a given target style combination, the style transfer computing system applies the style transfer language model on the source sentence to generate a target sentence in the given target style combination.

公开(公告)号：US20210406465A1

公开(公告)日：2021-12-30

申请号：US17467672

申请日：2021-09-07

Applicant: Adobe Inc.

Inventor： Balaji Vasan Srinivasan ,  Gaurav Verma ,  Bakhtiyar Hussain Syed ,  Anandhavelu Natarajan

IPC: G06F40/253 ,  G06F40/44 ,  G06F40/166

Abstract: Rewriting text in the writing style of a target author is described. A stylistic rewriting system receives input text and an indication of the target author. The system trains a language model to understand the target author's writing style using a corpus of text associated with the target author. The language model may be transformer-based, and is first trained on a different corpus of text associated with a range of different authors to understand linguistic nuances of a particular language. Copies of the language model are then cascaded into an encoder-decoder framework, which is further trained using a masked language modeling objective and a noisy version of the target author corpus. After training, the encoder-decoder framework of the trained language model automatically rewrites input text in the writing style of the target author and outputs the rewritten text as stylized text.

公开(公告)号：US10685050B2

公开(公告)日：2020-06-16

申请号：US15960505

申请日：2018-04-23

Applicant: Adobe Inc.

Inventor： Kundan Krishna ,  Balaji Vasan Srinivasan

IPC: G06F16/34 ,  G06F16/33 ,  G06N20/00 ,  G06F16/31

Abstract: A word generation model obtains textual content and a requested topic of interest, and generates a targeted summary of the textual content tuned to the topic of interest. To do so, a topic-aware encoding model encodes the textual content with a topic label corresponding to the topic of interest to generate topic-aware encoded text. A word generation model selects a next word for the topic-based summary from the topic-aware encoded text. The word generation model is trained to generate topic-based summaries using machine learning on training data including a multitude of documents, a respective summary of each document, and a respective topic of each summary. Feedback of the selected next word is provided to the word generation model. The feedback causes the word generation model to select subsequent words for the topic-based summary based on the feedback of the next selected word.