Zobrazeno 1 - 10
of 176 065
pro vyhledávání: '"P. Span"'
A span of a given graph $G$ is the maximum distance that two players can keep at all times while visiting all vertices (edges) of $G$ and moving according to certain rules, that produce different variants of span. We prove that the vertex and edge sp
Externí odkaz:
http://arxiv.org/abs/2410.19524
Autor:
Zhang, Lily H., Dadkhahi, Hamid, Finkelstein, Mara, Trabelsi, Firas, Luo, Jiaming, Freitag, Markus
Despite growing interest in incorporating feedback to improve language models, most efforts focus only on sequence-level annotations. In this work, we explore the potential of utilizing fine-grained span-level annotations from offline datasets to imp
Externí odkaz:
http://arxiv.org/abs/2410.16509
This paper proposes a computationally tractable algorithm for learning infinite-horizon average-reward linear mixture Markov decision processes (MDPs) under the Bellman optimality condition. Our algorithm for linear mixture MDPs achieves a nearly min
Externí odkaz:
http://arxiv.org/abs/2410.14992
Autor:
Li, Zixuan, Xiong, Jing, Ye, Fanghua, Zheng, Chuanyang, Wu, Xun, Lu, Jianqiao, Wan, Zhongwei, Liang, Xiaodan, Li, Chengming, Sun, Zhenan, Kong, Lingpeng, Wong, Ngai
We present UncertaintyRAG, a novel approach for long-context Retrieval-Augmented Generation (RAG) that utilizes Signal-to-Noise Ratio (SNR)-based span uncertainty to estimate similarity between text chunks. This span uncertainty enhances model calibr
Externí odkaz:
http://arxiv.org/abs/2410.02719
Ensembling various LLMs to unlock their complementary potential and leverage their individual strengths is highly valuable. Previous studies typically focus on two main paradigms: sample-level and token-level ensembles. Sample-level ensemble methods
Externí odkaz:
http://arxiv.org/abs/2409.18583
Autor:
Bekos, Michael, Da Lozzo, Giordano, Frati, Fabrizio, Gupta, Siddharth, Kindermann, Philipp, Liotta, Giuseppe, Rutter, Ignaz, Tollis, Ioannis G.
This paper studies planar drawings of graphs in which each vertex is represented as a point along a sequence of horizontal lines, called levels, and each edge is either a horizontal segment or a strictly $y$-monotone curve. A graph is $s$-span weakly
Externí odkaz:
http://arxiv.org/abs/2409.01889
Autor:
Lin, Jiayi, Zhang, Chenyang, Tong, Haibo, Zhang, Dongyu, Hong, Qingqing, Hou, Bingxuan, Wang, Junli
Multi-Span Question Answering (MSQA) requires models to extract one or multiple answer spans from a given context to answer a question. Prior work mainly focuses on designing specific methods or applying heuristic strategies to encourage models to pr
Externí odkaz:
http://arxiv.org/abs/2410.16788
Autor:
Benning, Felix, Döring, Leif
We prove that all 'gradient span algorithms' have asymptotically deterministic behavior on scaled Gaussian random functions as the dimension tends to infinity. In particular, this result explains the counterintuitive phenomenon that different trainin
Externí odkaz:
http://arxiv.org/abs/2410.09973
In aerodynamic design, the use of Computational Fluid Dynamics (CFD) tools has long relied on Reynolds-Averaged Navier-Stokes (RANS) and Unsteady RANS (URANS) turbulence models. However, these methods face limitations in predicting massively separate
Externí odkaz:
http://arxiv.org/abs/2410.01254
Although language model scores are often treated as probabilities, their reliability as probability estimators has mainly been studied through calibration, overlooking other aspects. In particular, it is unclear whether language models produce the sa
Externí odkaz:
http://arxiv.org/abs/2409.19984