Zobrazeno 1 - 10
of 37
pro vyhledávání: '"Byerly, Adam"'
Autor:
Byerly, Adam, Khashabi, Daniel
Self-consistency (SC) has been demonstrated to enhance the performance of large language models (LLMs) across various tasks and domains involving short content. However, does this evidence support its effectiveness for long-context problems? This stu
Externí odkaz:
http://arxiv.org/abs/2411.01101
Large Language Models (LLMs) exhibit positional bias, struggling to utilize information from the middle or end of long contexts. Our study explores LLMs' long-context reasoning by probing their hidden representations. We find that while LLMs encode t
Externí odkaz:
http://arxiv.org/abs/2406.14673
Autor:
Xu, Kevin, Kordi, Yeganeh, Nayak, Tanay, Asija, Ado, Wang, Yizhong, Sanders, Kate, Byerly, Adam, Zhang, Jingyu, Van Durme, Benjamin, Khashabi, Daniel
Can advanced multi-modal models effectively tackle complex web-based tasks? Such tasks are often found on crowdsourcing platforms, where crowdworkers engage in challenging micro-tasks within web-based environments. Building on this idea, we present T
Externí odkaz:
http://arxiv.org/abs/2403.11905
Autor:
Byerly, Adam, Kalganova, Tatiana
We provide a definition for class density that can be used to measure the aggregate similarity of the samples within each of the classes in a high-dimensional, unstructured dataset. We then put forth several candidate methods for calculating class de
Externí odkaz:
http://arxiv.org/abs/2202.03856
Autor:
Byerly, Adam, Kalganova, Tatiana
We show that, for each of five datasets of increasing complexity, certain training samples are more informative of class membership than others. These samples can be identified a priori to training by analyzing their position in reduced dimensional s
Externí odkaz:
http://arxiv.org/abs/2202.03238
We present a dataset consisting of high-resolution images of 13 micro-PCBs captured in various rotations and perspectives relative to the camera, with each sample labeled for PCB type, rotation category, and perspective categories. We then present th
Externí odkaz:
http://arxiv.org/abs/2101.11164
Most capsule network designs rely on traditional matrix multiplication between capsule layers and computationally expensive routing mechanisms to deal with the capsule dimensional entanglement that the matrix multiplication introduces. By using Homog
Externí odkaz:
http://arxiv.org/abs/2001.09136
Autor:
Byerly, Adam, Kalganova, Tatiana
Capsules are the name given by Geoffrey Hinton to vector-valued neurons. Neural networks traditionally produce a scalar value for an activated neuron. Capsules, on the other hand, produce a vector of values, which Hinton argues correspond to a single
Externí odkaz:
http://arxiv.org/abs/1906.08676
Publikováno v:
In Neurocomputing 6 November 2021 463:545-553
Akademický článek
Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.
K zobrazení výsledku je třeba se přihlásit.