DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | 이주호 | - |
dc.contributor.author | Moon, Taehong | - |
dc.contributor.author | 문태홍 | - |
dc.date.accessioned | 2024-07-25T19:30:45Z | - |
dc.date.available | 2024-07-25T19:30:45Z | - |
dc.date.issued | 2023 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1045724&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/320536 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 김재철AI대학원, 2023.8,[iii, 18 p. :] | - |
dc.description.abstract | Diffusion models have shown remarkable performance in generation problems over various domains including images, videos, text, and audio. A practical bottleneck of diffusion models is their sampling speed, due to the repeated evaluation of score estimation networks during the inference. In this work, we propose a novel framework capable of adaptively allocating compute required for the score estimation, thereby reducing the overall sampling time of diffusion models. We observe that the amount of computation required for the score estimation may vary along the time step for which the score is estimated. Based on this observation, we propose an early-exiting scheme, where we skip the subset of parameters in the score estimation network during the inference, based on a time-dependent exit schedule. Using the diffusion models for image synthesis, we show that our method could significantly improve the sampling throughput of the diffusion models without compromising image quality. Furthermore, we also demonstrate that our method seamlessly integrates with various types of solvers for faster sampling, capitalizing on their compatibility to enhance overall efficiency. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | 디퓨전 모델▼a빠른 추론▼a조기 종료 | - |
dc.subject | Diffusion models▼aAccelerated inference▼aEarly exiting | - |
dc.title | Adaptive score estimation: early exiting for accelerated inference in diffusion models | - |
dc.title.alternative | 디퓨전 모델의 빠른 추론을 위한 조기 종료 알고리즘 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :김재철AI대학원, | - |
dc.contributor.alternativeauthor | Lee, Juho | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.