Zobrazeno 1 - 8
of 8
pro vyhledávání: '"Karpas, Ehud"'
Autor:
Bercovich, Akhiad, Ronen, Tomer, Abramovich, Talor, Ailon, Nir, Assaf, Nave, Dabbah, Mohammad, Galil, Ido, Geifman, Amnon, Geifman, Yonatan, Golan, Izhak, Haber, Netanel, Karpas, Ehud, Levy, Itay, Mor, Shahar, Moshe, Zach, Nabwani, Najeeb, Puny, Omri, Rubin, Ran, Schen, Itamar, Shahaf, Ido, Tropp, Oren, Argov, Omer Ullman, Zilberstein, Ran, El-Yaniv, Ran
Large language models (LLMs) have demonstrated remarkable capabilities, but their adoption is limited by high computational costs during inference. While increasing parameter counts enhances accuracy, it also widens the gap between state-of-the-art c
Externí odkaz:
http://arxiv.org/abs/2411.19146
Autor:
Ratner, Nir, Levine, Yoav, Belinkov, Yonatan, Ram, Ori, Magar, Inbal, Abend, Omri, Karpas, Ehud, Shashua, Amnon, Leyton-Brown, Kevin, Shoham, Yoav
When applied to processing long text, Large Language Models (LLMs) are limited by their context window. Existing efforts to address this limitation involve training specialized architectures, and cannot be easily applied to off-the-shelf LLMs. We pre
Externí odkaz:
http://arxiv.org/abs/2212.10947
Autor:
Karpas, Ehud, Abend, Omri, Belinkov, Yonatan, Lenz, Barak, Lieber, Opher, Ratner, Nir, Shoham, Yoav, Bata, Hofit, Levine, Yoav, Leyton-Brown, Kevin, Muhlgay, Dor, Rozen, Noam, Schwartz, Erez, Shachaf, Gal, Shalev-Shwartz, Shai, Shashua, Amnon, Tenenholtz, Moshe
Huge language models (LMs) have ushered in a new era for AI, serving as a gateway to natural-language-based knowledge tasks. Although an essential element of modern AI, LMs are also inherently limited in a number of ways. We discuss these limitations
Externí odkaz:
http://arxiv.org/abs/2205.00445
Autor:
Ginosar, Gily1 (AUTHOR), Karpas, Ehud D.1 (AUTHOR), Weitzner, Idan2 (AUTHOR), Ulanovsky, Nachum1 (AUTHOR) nachum.ulanovsky@weizmann.ac.il
Publikováno v:
Scientific Reports. 7/12/2023, Vol. 13 Issue 1, p1-13. 13p.
Autor:
Karpas, Ehud D., Kun, Ferenc
Publikováno v:
Physical Review E 93, 033002 (2016)
Quasi-brittle behavior where macroscopic failure is preceded by stable damaging and intensive cracking activity is a desired feature of materials because it makes fracture predictable. Based on a fiber bundle model with global load sharing we show th
Externí odkaz:
http://arxiv.org/abs/1604.01430
Autor:
Karpas, Ehud, Kun, Ferenc
Publikováno v:
Europhysics Letters 95, 16004 (2011)
We investigate the fracture process of a bundle of fibers with random Young modulus and a constant breaking strength. For two component systems we show that the strength of the mixture is always lower than the strength of the individual components. F
Externí odkaz:
http://arxiv.org/abs/1106.4748
Publikováno v:
Proceedings of the National Academy of Sciences of the United States of America, 2017 May . 114(22), 5589-5594.
Externí odkaz:
https://www.jstor.org/stable/26483356