Zobrazeno 1 - 10
of 190
pro vyhledávání: '"Crisostomi, P."'
Autor:
Gargiulo, Antonio Andrea, Crisostomi, Donato, Bucarelli, Maria Sofia, Scardapane, Simone, Silvestri, Fabrizio, Rodolà, Emanuele
Task Arithmetic has emerged as a simple yet effective method to merge models without additional training. However, by treating entire networks as flat parameter vectors, it overlooks key structural information and is susceptible to task interference.
Externí odkaz:
http://arxiv.org/abs/2412.00081
Autor:
Zhou, Luca, Solombrino, Daniele, Crisostomi, Donato, Bucarelli, Maria Sofia, Silvestri, Fabrizio, Rodolà, Emanuele
Model merging has recently emerged as a cost-efficient paradigm for multi-task learning. Among current approaches, task arithmetic stands out for its simplicity and effectiveness. In this paper, we motivate the effectiveness of task vectors by linkin
Externí odkaz:
http://arxiv.org/abs/2411.03055
The green potential of electric vehicles (EVs) can be fully realized only if their batteries are charged using energy generated from renewable (i.e. green) sources. For logistic or economic reasons, however, EV drivers may be tempted to avoid chargin
Externí odkaz:
http://arxiv.org/abs/2410.18971
Autor:
Wieberneit, Felix, Crisostomi, Emanuele, Quinn, Anthony, Hamedmoghadam, Homayoun, Ferraro, Pietro, Shorten, Robert
In this paper, we introduce a quantitative framework to optimize electric vehicle (EV) battery capacities, considering two criteria: upfront vehicle cost and charging inconvenience cost. For this purpose, we (1) develop a comprehensive model for char
Externí odkaz:
http://arxiv.org/abs/2410.16997
Numerical simulations of the Cauchy problem for self-interacting massive vector fields often face instabilities and apparent pathologies. We explicitly demonstrate that these issues, previously reported in the literature, are actually due to the brea
Externí odkaz:
http://arxiv.org/abs/2407.08774
In this paper, we present a novel data-free method for merging neural networks in weight space. Differently from most existing works, our method optimizes for the permutations of network neurons globally across all layers. This allows us to enforce c
Externí odkaz:
http://arxiv.org/abs/2405.17897
In the gravitational-wave analysis of pulsar-timing-array datasets, parameter estimation is usually performed using Markov Chain Monte Carlo methods to explore posterior probability densities. We introduce an alternative procedure that relies instead
Externí odkaz:
http://arxiv.org/abs/2405.08857
Publikováno v:
PhysRevD.110.123007(2024)
We propose a novel method ($floZ$), based on normalizing flows, to estimate the Bayesian evidence (and its numerical uncertainty) from a pre-existing set of samples drawn from the unnormalized posterior distribution. We validate it on distributions w
Externí odkaz:
http://arxiv.org/abs/2404.12294
Autor:
Crisostomi, Donato, Cannistraci, Irene, Moschella, Luca, Barbiero, Pietro, Ciccone, Marco, Liò, Pietro, Rodolà, Emanuele
Models trained on semantically related datasets and tasks exhibit comparable inter-sample relations within their latent spaces. We investigate in this study the aggregation of such latent spaces to create a unified space encompassing the combined inf
Externí odkaz:
http://arxiv.org/abs/2311.06547
In this work, we introduce an efficient generation procedure to produce synthetic multi-modal datasets of fluid simulations. The procedure can reproduce the dynamics of fluid flows and allows for exploring and learning various properties of their com
Externí odkaz:
http://arxiv.org/abs/2311.06284