Zobrazeno 1 - 10
of 571
pro vyhledávání: '"Metel, A."'
We present a simple on the fly method for faster inference of large language models. Unlike other (self-)speculative decoding techniques, our method does not require fine-tuning or black-box optimization to generate a fixed draft model, relying inste
Externí odkaz:
http://arxiv.org/abs/2410.01028
Autor:
Sergey N. Grigoriev, Alexander S. Metel, Marina A. Volosova, Enver S. Mustafaev, Yury A. Melnik
Publikováno v:
Surfaces, Vol 7, Iss 3, Pp 714-724 (2024)
The surfaces of ceramic products are replete with numerous defects, such as those that appear during the diamond grinding of sintered SiAlON ceramics. The defective surface layer is the reason for the low effectiveness of TiZrN coatings under abrasiv
Externí odkaz:
https://doaj.org/article/25f042e10e604919acd7ad83e597d571
Autor:
Nia, Vahid Partovi, Zhang, Guojun, Kobyzev, Ivan, Metel, Michael R., Li, Xinlin, Sun, Ke, Hemati, Sobhan, Asgharian, Masoud, Kong, Linglong, Liu, Wulong, Chen, Boxing
Deep models are dominating the artificial intelligence (AI) industry since the ImageNet challenge in 2012. The size of deep models is increasing ever since, which brings new challenges to this field with applications in cell phones, personal computer
Externí odkaz:
http://arxiv.org/abs/2303.15464
Autor:
Metel, Michael R.
Motivated by neural network training in low-precision arithmetic environments, this work studies the convergence of variants of SGD using adaptive step sizes with computational error. Considering a general stochastic Lipschitz continuous loss functio
Externí odkaz:
http://arxiv.org/abs/2211.04655
Autor:
Alexander S. Metel, Marina A. Volosova, Enver S. Mustafaev, Yury A. Melnik, Anna A. Okunkova, Sergey N. Grigoriev
Publikováno v:
Ceramics, Vol 7, Iss 1, Pp 55-67 (2024)
The surface of ceramic products manufactured using diamond grinding is replete with shallow scratches, deep grooves and other defects. The thickness of the defective layer amounts to 3–4 µm and it must be removed to increase wear resistance of the
Externí odkaz:
https://doaj.org/article/7abcb8e4ec804786a5cbc07df08e141e
Autor:
Metel, Michael R.
This paper is motivated by structured sparsity for deep neural network training. We study a weighted group L0-norm constraint, and present the projection and normal cone of this set. Using randomized smoothing, we develop zeroth and first-order algor
Externí odkaz:
http://arxiv.org/abs/2202.06141
Autor:
Grigoriev, Sergey N.1 (AUTHOR) s.grigoriev@stankin.ru, Metel, Alexander S.1 (AUTHOR) a.metel@stankin.ru, Volosova, Marina A.1 (AUTHOR), Mustafaev, Enver S.1 (AUTHOR), Melnik, Yury A.1 (AUTHOR)
Publikováno v:
Surfaces (2571-9637). Sep2024, Vol. 7 Issue 3, p714-724. 11p.
Autor:
Olga Metel
Publikováno v:
Quaestio Rossica, Vol 12, Iss 2 (2024)
Представлен анализ сборника интервью и воспоминаний сотрудников Института российской истории РАН «Ты идешь в храм науки…», подготовле
Externí odkaz:
https://doaj.org/article/7b931c28fad04c0b85cb05772cb44622
Autor:
Alexander Metel, Alexey Vereschaka, Catherine Sotova, Anton Seleznev, Nikolay Sitnikov, Filipp Milovich, Kirill Makarevich, Sergey Grigoriev
Publikováno v:
Technologies, Vol 12, Iss 10, p 179 (2024)
The fracture strength was compared in a scratch test of coatings based on the ZrN system with the introduction of Ti, Nb and Hf, which were deposited on a titanium alloy substrate. The coatings were deposited using Controlled Accelerated Arc (CAA-PVD
Externí odkaz:
https://doaj.org/article/38eea21ce80f4d938025f014ef11ce11
Autor:
Metel, Michael R., Takeda, Akiko
This paper considers a general convex constrained problem setting where functions are not assumed to be differentiable nor Lipschitz continuous. Our motivation is in finding a simple first-order method for solving a wide range of convex optimization
Externí odkaz:
http://arxiv.org/abs/2009.12769