Zobrazeno 1 - 10
of 82 232
pro vyhledávání: '"Joseph, E"'
Autor:
Krentsel, Alexander, Schafhalter, Peter, Gonzalez, Joseph E., Ratnasamy, Sylvia, Shenker, Scott, Stoica, Ion
Prevailing wisdom asserts that one cannot rely on the cloud for critical real-time control systems like self-driving cars. We argue that we can, and must. Following the trends of increasing model sizes, improvements in hardware, and evolving mobile n
Externí odkaz:
http://arxiv.org/abs/2410.16227
Autor:
Frick, Evan, Li, Tianle, Chen, Connor, Chiang, Wei-Lin, Angelopoulos, Anastasios N., Jiao, Jiantao, Zhu, Banghua, Gonzalez, Joseph E., Stoica, Ion
We introduce a new benchmark for reward models that quantifies their ability to produce strong language models through RLHF (Reinforcement Learning from Human Feedback). The gold-standard approach is to run a full RLHF training pipeline and directly
Externí odkaz:
http://arxiv.org/abs/2410.14872
This project, named HEnRY, aims to introduce a Multi-Agent System (MAS) into Intesa Sanpaolo. The name HEnRY summarizes the project's core principles: the Hierarchical organization of agents in a layered structure for efficient resource management; E
Externí odkaz:
http://arxiv.org/abs/2410.12720
Generating diverse responses from large language models (LLMs) is crucial for applications such as planning/search and synthetic data generation, where diversity provides distinct answers across generations. Prior approaches rely on increasing temper
Externí odkaz:
http://arxiv.org/abs/2410.09038
Autor:
Yang, Ling, Yu, Zhaochen, Zhang, Tianjun, Xu, Minkai, Gonzalez, Joseph E., Cui, Bin, Yan, Shuicheng
Large language models (LLMs) like GPT-4, PaLM, and LLaMA have shown significant improvements in various reasoning tasks. However, smaller models such as Llama-3-8B and DeepSeekMath-Base still struggle with complex mathematical reasoning because they
Externí odkaz:
http://arxiv.org/abs/2410.09008
Large language models (LLMs) often exhibit subtle yet distinctive characteristics in their outputs that users intuitively recognize, but struggle to quantify. These "vibes" -- such as tone, formatting, or writing style -- influence user preferences,
Externí odkaz:
http://arxiv.org/abs/2410.12851
We present a novel semiclassical phase-space surface hopping approach that goes beyond the Born-Oppenheimer approximation and all existing surface hopping formalisms. We demonstrate that working with a correct phase-space electronic Hamiltonian can c
Externí odkaz:
http://arxiv.org/abs/2410.01156
Autor:
Qiu, Tian, Subotnik, Joseph E.
We propose an efficient algorithm for the recently published electron/hole-transfer Dynamical-weighted State-averaged Constrained CASSCF (eDSC/hDSC) method studying charge transfer states and D$_1$-D$_0$ crossings for systems with odd numbers of elec
Externí odkaz:
http://arxiv.org/abs/2409.14631
The Automated Audio Captioning (AAC) task asks models to generate natural language descriptions of an audio input. Evaluating these machine-generated audio captions is a complex task that requires considering diverse factors, among them, auditory sce
Externí odkaz:
http://arxiv.org/abs/2409.12962
Autor:
Thygesen, Erica, Rodriguez, Joseph E., De Beurs, Zoë L., Vanderburg, Andrew, Livingston, John H., Irwin, Jonathon, Venner, Alexander, Cretignier, Michael, Collins, Karen A., Bieryla, Allyson, Charbonneau, David, Crossfield, Ian J. M., Dumusque, Xavier, Kielkopf, John, Latham, David W., Werner, Michael
K2-2 b/HIP 116454 b, the first exoplanet discovery by K2 during its Two-Wheeled Concept Engineering Test, is a sub-Neptune (2.5 $\pm$ 0.1 $R_\oplus$, 9.7 $\pm$ 1.2 $M_\oplus$) orbiting a relatively bright (KS = 8.03) K-dwarf on a 9.1 day period. Unfo
Externí odkaz:
http://arxiv.org/abs/2409.07019