Zobrazeno 1 - 10
of 112
pro vyhledávání: '"Mroueh, Youssef"'
Autor:
Ko, Ching-Yun, Chen, Pin-Yu, Das, Payel, Mroueh, Youssef, Dan, Soham, Kollias, Georgios, Chaudhury, Subhajit, Pedapati, Tejaswini, Daniel, Luca
Reducing the likelihood of generating harmful and toxic output is an essential task when aligning large language models (LLMs). Existing methods mainly rely on training an external reward model (i.e., another language model) or fine-tuning the LLM us
Externí odkaz:
http://arxiv.org/abs/2410.03818
Autor:
Zhang, Zhengxin, Goldfeld, Ziv, Greenewald, Kristjan, Mroueh, Youssef, Sriperumbudur, Bharath K.
The Wasserstein space of probability measures is known for its intricate Riemannian structure, which underpins the Wasserstein geometry and enables gradient flow algorithms. However, the Wasserstein geometry may not be suitable for certain tasks or d
Externí odkaz:
http://arxiv.org/abs/2407.11800
Stochastic dominance is an important concept in probability theory, econometrics and social choice theory for robustly modeling agents' preferences between random outcomes. While many works have been dedicated to the univariate case, little has been
Externí odkaz:
http://arxiv.org/abs/2406.06425
Autor:
Mroueh, Youssef
Policy alignment of large language models refers to constrained policy optimization, where the policy is optimized to maximize a reward while staying close to a reference policy with respect to an $f$-divergence such as the $\mathsf{KL}$ divergence.
Externí odkaz:
http://arxiv.org/abs/2406.05883
Autor:
Melnyk, Igor, Mroueh, Youssef, Belgodere, Brian, Rigotti, Mattia, Nitsure, Apoorva, Yurochkin, Mikhail, Greenewald, Kristjan, Navratil, Jiri, Ross, Jerret
Current LLM alignment techniques use pairwise human preferences at a sample level, and as such, they do not imply an alignment on the distributional level. We propose in this paper Alignment via Optimal Transport (AOT), a novel method for distributio
Externí odkaz:
http://arxiv.org/abs/2406.05882
Autor:
Ross, Jerret, Belgodere, Brian, Hoffman, Samuel C., Chenthamarakshan, Vijil, Mroueh, Youssef, Das, Payel
Transformer-based models trained on large and general purpose datasets consisting of molecular strings have recently emerged as a powerful tool for successfully modeling various structure-property relations. Inspired by this success, we extend the pa
Externí odkaz:
http://arxiv.org/abs/2405.04912
Autor:
Nitsure, Apoorva, Mroueh, Youssef, Rigotti, Mattia, Greenewald, Kristjan, Belgodere, Brian, Yurochkin, Mikhail, Navratil, Jiri, Melnyk, Igor, Ross, Jerret
We propose a distributional framework for benchmarking socio-technical risks of foundation models with quantified statistical significance. Our approach hinges on a new statistical relative testing based on first and second order stochastic dominance
Externí odkaz:
http://arxiv.org/abs/2310.07132
Autor:
Belgodere, Brian, Dognin, Pierre, Ivankay, Adam, Melnyk, Igor, Mroueh, Youssef, Mojsilovic, Aleksandra, Navratil, Jiri, Nitsure, Apoorva, Padhi, Inkit, Rigotti, Mattia, Ross, Jerret, Schiff, Yair, Vedpathak, Radhika, Young, Richard A.
Real-world data often exhibits bias, imbalance, and privacy risks. Synthetic datasets have emerged to address these issues. This paradigm relies on generative AI models to generate unbiased, privacy-preserving data while maintaining fidelity to the o
Externí odkaz:
http://arxiv.org/abs/2304.10819
The Gromov-Wasserstein (GW) distance, rooted in optimal transport (OT) theory, quantifies dissimilarity between metric measure spaces and provides a framework for aligning heterogeneous datasets. While computational aspects of the GW problem have bee
Externí odkaz:
http://arxiv.org/abs/2212.12848
Generative adversarial networks (GANs) are a class of machine-learning models that use adversarial training to generate new samples with the same (potentially very complex) statistics as the training samples. One major form of training failure, known
Externí odkaz:
http://arxiv.org/abs/2212.04580