Zobrazeno 1 - 10
of 4 089
pro vyhledávání: '"Ermis, A."'
The use of synthetic data has played a critical role in recent state-of-art breakthroughs. However, overly relying on a single oracle teacher model to generate data has been shown to lead to model collapse and invite propagation of biases. These limi
Externí odkaz:
http://arxiv.org/abs/2408.14960
We present a tight inequality to test the dynamical nature of spacetime. A general-relativistic violation of that inequality certifies change of curvature, in the same sense as a quantum-mechanical violation of a Bell inequality certifies a source of
Externí odkaz:
http://arxiv.org/abs/2407.17203
Autor:
Aakanksha, Ahmadian, Arash, Ermis, Beyza, Goldfarb-Tarrant, Seraphina, Kreutzer, Julia, Fadaee, Marzieh, Hooker, Sara
A key concern with the concept of "alignment" is the implicit question of "alignment to what?". AI systems are increasingly used across the world, yet safety alignment is often focused on homogeneous monolingual settings. Additionally, preference tra
Externí odkaz:
http://arxiv.org/abs/2406.18682
Online platforms generate hundreds of billions of dollars in revenue per year by showing advertisements alongside their own content. Currently, these platforms are integrating Large Language Models (LLMs) into their services. This makes revenue gener
Externí odkaz:
http://arxiv.org/abs/2405.05905
Autor:
Zhang, Jingwei, Swinnen, Lauren, Chatzichristos, Christos, Broux, Victoria, Proost, Renee, Jansen, Katrien, Mahler, Benno, Zabler, Nicolas, Epitashvilli, Nino, Dümpelmann, Matthias, Schulze-Bonhage, Andreas, Schriewer, Elisabeth, Ermis, Ummahan, Wolking, Stefan, Linke, Florian, Weber, Yvonne, Symmonds, Mkael, Sen, Arjune, Biondi, Andrea, Richardson, Mark P., Sulaiman I, Abuhaiba, Silva, Ana Isabel, Sales, Francisco, Vértes, Gergely, Van Paesschen, Wim, De Vos, Maarten
Objective: Most current wearable tonic-clonic seizure (TCS) detection systems are based on extra-cerebral signals, such as electromyography (EMG) or accelerometry (ACC). Although many of these devices show good sensitivity in seizure detection, their
Externí odkaz:
http://arxiv.org/abs/2403.13066
To date, toxicity mitigation in language models has almost entirely been focused on single-language settings. As language models embrace multilingual capabilities, it's crucial our safety measures keep pace. Recognizing this research gap, our approac
Externí odkaz:
http://arxiv.org/abs/2403.03893
Autor:
Yıldız, Çağatay, Ravichandran, Nishaanth Kanna, Punia, Prishruit, Bethge, Matthias, Ermis, Beyza
This paper studies the evolving domain of Continual Learning (CL) in large language models (LLMs), with a focus on developing strategies for efficient and sustainable training. Our primary emphasis is on continual domain-adaptive pretraining, a proce
Externí odkaz:
http://arxiv.org/abs/2402.17400
Human evaluation is increasingly critical for assessing large language models, capturing linguistic nuances, and reflecting user preferences more accurately than traditional automated metrics. However, the resource-intensive nature of this type of an
Externí odkaz:
http://arxiv.org/abs/2310.14424
Considerable effort has been dedicated to mitigating toxicity, but existing methods often require drastic modifications to model parameters or the use of computationally intensive auxiliary models. Furthermore, previous approaches have often neglecte
Externí odkaz:
http://arxiv.org/abs/2310.07589
We derive multiparty games that, if the winning chance exceeds a certain limit, prove the incompatibility of the parties' causal relations with any partial order. This, in turn, means that the parties exert a back-action on the causal relations; the
Externí odkaz:
http://arxiv.org/abs/2309.15752