Zobrazeno 1 - 10
of 815
pro vyhledávání: '"Günther Michael"'
Autor:
Sturua, Saba, Mohr, Isabelle, Akram, Mohammad Kalim, Günther, Michael, Wang, Bo, Krimmel, Markus, Wang, Feng, Mastrapas, Georgios, Koukounas, Andreas, Wang, Nan, Xiao, Han
We introduce jina-embeddings-v3, a novel text embedding model with 570 million parameters, achieves state-of-the-art performance on multilingual data and long-context retrieval tasks, supporting context lengths of up to 8192 tokens. The model include
Externí odkaz:
http://arxiv.org/abs/2409.10173
Many use cases require retrieving smaller portions of text, and dense vector-based retrieval systems often perform better with shorter text segments, as the semantics are less likely to be over-compressed in the embeddings. Consequently, practitioner
Externí odkaz:
http://arxiv.org/abs/2409.04701
Autor:
Jha, Rohan, Wang, Bo, Günther, Michael, Mastrapas, Georgios, Sturua, Saba, Mohr, Isabelle, Koukounas, Andreas, Akram, Mohammad Kalim, Wang, Nan, Xiao, Han
Multi-vector dense models, such as ColBERT, have proven highly effective in information retrieval. ColBERT's late interaction scoring approximates the joint query-document attention seen in cross-encoders while maintaining inference efficiency closer
Externí odkaz:
http://arxiv.org/abs/2408.16672
Autor:
Zaspel, Peter, Günther, Michael
Port-Hamiltonian systems (pHS) allow for a structure-preserving modeling of dynamical systems. Coupling pHS via linear relations between input and output defines an overall pHS, which is structure preserving. However, in multiphysics applications, so
Externí odkaz:
http://arxiv.org/abs/2406.18726
Operator splitting methods tailored to coupled linear port-Hamiltonian systems are developed. We present algorithms that are able to exploit scalar coupling, as well as multirate potential of these coupled systems. The obtained algorithms preserve th
Externí odkaz:
http://arxiv.org/abs/2406.17311
We propose a modification to the Cayley transform that defines a suitable local parameterization for the special unitary group $\mathrm{SU(3)}$. The new mapping is used to construct splitting methods for separable Hamiltonian systems whose phase spac
Externí odkaz:
http://arxiv.org/abs/2406.11337
Autor:
Koukounas, Andreas, Mastrapas, Georgios, Günther, Michael, Wang, Bo, Martens, Scott, Mohr, Isabelle, Sturua, Saba, Akram, Mohammad Kalim, Martínez, Joan Fontanals, Ognawala, Saahil, Guzman, Susana, Werk, Maximilian, Wang, Nan, Xiao, Han
Contrastive Language-Image Pretraining (CLIP) is widely used to train models to align images and texts in a common embedding space by mapping them to fixed-sized vectors. These models are key to multimodal information retrieval and related tasks. How
Externí odkaz:
http://arxiv.org/abs/2405.20204
Electromagneto-quasistatic (EMQS) field formulations are often dubbed as Darwin-type field formulations which approximate the Maxwell equations by neglecting radiation effects while modelling resistive, capacitive, and inductive effects. A common fea
Externí odkaz:
http://arxiv.org/abs/2404.18767
In the simulation of differential-algebraic equations (DAEs), it is essential to employ numerical schemes that take into account the inherent structure and maintain explicit or hidden algebraic constraints without altering them. This paper focuses on
Externí odkaz:
http://arxiv.org/abs/2404.13531
We propose a new framework of Hessian-free force-gradient integrators that do not require the analytical expression of the force-gradient term based on the Hessian of the potential. Due to that the new class of decomposition algorithms for separable
Externí odkaz:
http://arxiv.org/abs/2403.10370