Zobrazeno 1 - 10
of 528
pro vyhledávání: '"ERMON, STEFANO"'
We consider the training of the first layer of vision models and notice the clear relationship between pixel values and gradient update magnitudes: the gradients arriving at the weights of a first layer are by definition directly proportional to (nor
Externí odkaz:
http://arxiv.org/abs/2410.23970
Preference optimization has made significant progress recently, with numerous methods developed to align language models with human preferences. This paper introduces $f$-divergence Preference Optimization ($f$-PO), a novel framework that generalizes
Externí odkaz:
http://arxiv.org/abs/2410.21662
Autor:
Xu, Minkai, Geffner, Tomas, Kreis, Karsten, Nie, Weili, Xu, Yilun, Leskovec, Jure, Ermon, Stefano, Vahdat, Arash
Despite remarkable progress in autoregressive language models, alternative generative paradigms beyond left-to-right generation are still being actively explored. Discrete diffusion models, with the capacity for parallel generation, have recently eme
Externí odkaz:
http://arxiv.org/abs/2410.21357
Synthesizing high-quality tabular data is an important topic in many data science tasks, ranging from dataset augmentation to privacy protection. However, developing expressive generative models for tabular data is challenging due to its inherent het
Externí odkaz:
http://arxiv.org/abs/2410.20626
Autor:
Petersen, Felix, Borgelt, Christian, Sutter, Tobias, Kuehne, Hilde, Deussen, Oliver, Ermon, Stefano
When training neural networks with custom objectives, such as ranking losses and shortest-path losses, a common problem is that they are, per se, non-differentiable. A popular approach is to continuously relax the objectives to provide gradients, ena
Externí odkaz:
http://arxiv.org/abs/2410.19055
Autor:
Nguyen, Bac, Lai, and Chieh-Hsin, Takida, Yuhta, Murata, Naoki, Uesaka, Toshimitsu, Ermon, Stefano, Mitsufuji, Yuki
Latent diffusion models have enabled continuous-state diffusion models to handle a variety of datasets, including categorical data. However, most methods rely on fixed pretrained embeddings, limiting the benefits of joint training with the diffusion
Externí odkaz:
http://arxiv.org/abs/2410.14758
Generative models have shown great promise in generating 3D geometric systems, which is a fundamental problem in many natural science domains such as molecule and protein design. However, existing approaches only operate on static structures, neglect
Externí odkaz:
http://arxiv.org/abs/2410.13027
We deal with the problem of gradient estimation for stochastic differentiable relaxations of algorithms, operators, simulators, and other non-differentiable functions. Stochastic smoothing conventionally perturbs the input of a non-differentiable fun
Externí odkaz:
http://arxiv.org/abs/2410.08125
Autor:
Murata, Naoki, Lai, Chieh-Hsin, Takida, Yuhta, Uesaka, Toshimitsu, Nguyen, Bac, Ermon, Stefano, Mitsufuji, Yuki
Recent literature has effectively utilized diffusion models trained on continuous variables as priors for solving inverse problems. Notably, discrete diffusion models with discrete latent codes have shown strong performance, particularly in modalitie
Externí odkaz:
http://arxiv.org/abs/2410.14710
Autor:
Zeng, Bohan, Yang, Ling, Li, Siyu, Liu, Jiaming, Zhang, Zixiang, Tian, Juanxi, Zhu, Kaixin, Guo, Yongzhen, Wang, Fu-Yun, Xu, Minkai, Ermon, Stefano, Zhang, Wentao
Recent advances in diffusion models have demonstrated exceptional capabilities in image and video generation, further improving the effectiveness of 4D synthesis. Existing 4D generation methods can generate high-quality 4D objects or scenes based on
Externí odkaz:
http://arxiv.org/abs/2410.07155