Zobrazeno 1 - 10
of 2 024
pro vyhledávání: '"Hahn, Michael"'
We present revised point-spread functions (PSFs) for the Atmospheric Imaging Assembly (AIA) onboard the Solar Dynamics Observatory (SDO). These PSFs provide a robust estimate of the light diffracted by the meshes holding the entrance and focal plane
Externí odkaz:
http://arxiv.org/abs/2410.08967
Autor:
Huang, Xinting, Yang, Andy, Bhattamishra, Satwik, Sarrof, Yash, Krebs, Andreas, Zhou, Hattie, Nakkiran, Preetum, Hahn, Michael
A major challenge for transformers is generalizing to sequences longer than those observed during training. While previous works have empirically shown that transformers can either succeed or fail at length generalization depending on the task, theor
Externí odkaz:
http://arxiv.org/abs/2410.02140
Autor:
Fairchild, Alexander J., Hell, Natalie, Beiersdorfer, Peter, Brown, Gregory V., Eckart, Megan E., Hahn, Michael, Savin, Daniel W.
Solar physicists routinely utilize observations of Ar-like Fe IX and Cl-like Fe X emission to study a variety of solar structures. However, unidentified lines exist in the Fe IX and Fe X spectra, greatly impeding the spectroscopic diagnostic potentia
Externí odkaz:
http://arxiv.org/abs/2408.14454
Autor:
Hahn, Michael
Publikováno v:
Transactions of the Association for Computational Linguistics, Vol 8, Pp 156-171 (2020)
Transformers are emerging as the new workhorse of NLP, showing great success across tasks. Unlike LSTMs, transformers process input sequences entirely through self-attention. Previous work has suggested that the computational capabilities of self-att
Externí odkaz:
https://doaj.org/article/9a179321a285476ca44f605eba696e3e
Autor:
Hahn, Michael, Baroni, Marco
Publikováno v:
Transactions of the Association for Computational Linguistics, Vol 7, Pp 467-484 (2019)
Recurrent neural networks (RNNs) have reached striking performance in many natural language processing tasks. This has renewed interest in whether these generic sequence processing devices are inducing genuine linguistic knowledge. Nearly all current
Externí odkaz:
https://doaj.org/article/a88791791a77479eb6504d2852a10230
Transformer architectures have been widely adopted in foundation models. Due to their high inference costs, there is renewed interest in exploring the potential of efficient recurrent architectures (RNNs). In this paper, we analyze the differences in
Externí odkaz:
http://arxiv.org/abs/2406.09347
Recently, recurrent models based on linear state space models (SSMs) have shown promising performance in language modeling (LM), competititve with transformers. However, there is little understanding of the in-principle abilities of such models, whic
Externí odkaz:
http://arxiv.org/abs/2405.17394
The inner workings of neural networks can be better understood if we can fully decipher the information encoded in neural activations. In this paper, we argue that this information is embodied by the subset of inputs that give rise to similar activat
Externí odkaz:
http://arxiv.org/abs/2405.17653
Autor:
Futrell, Richard, Hahn, Michael
Human language is a unique form of communication in the natural world, distinguished by its structured nature. Most fundamentally, it is systematic, meaning that signals can be broken down into component parts that are individually meaningful -- roug
Externí odkaz:
http://arxiv.org/abs/2405.12109
Autor:
Hahn, Michael, Rofin, Mark
Empirical studies have identified a range of learnability biases and limitations of transformers, such as a persistent difficulty in learning to compute simple formal languages such as PARITY, and a bias towards low-degree functions. However, theoret
Externí odkaz:
http://arxiv.org/abs/2402.09963