Zobrazeno 1 - 10
of 11 203
pro vyhledávání: '"Ton That, P."'
Large Language Models (LLMs) have shown impressive performance in complex reasoning tasks through Chain-of-Thought (CoT) reasoning, allowing models to break down problems into manageable sub-tasks. However, existing CoT evaluation techniques either r
Externí odkaz:
http://arxiv.org/abs/2411.11984
The Bradley-Terry (BT) model is a common and successful practice in reward modeling for Large Language Model (LLM) alignment. However, it remains unclear why this model -- originally developed for multi-player stochastic game matching -- can be adopt
Externí odkaz:
http://arxiv.org/abs/2411.04991
Autor:
Bao, Nguyen Tran Gia, Trang, Ton Nu Quynh, Thoai, Nam, Thang, Phan Bach, Thu, Vu Thi Hanh, Hung, Nguyen Tuan
Direct Z-scheme heterostructures with enhanced redox potential are increasingly regarded as promising materials for solar-driven water splitting. This potential arises from the synergistic interaction between the intrinsic dipoles in Janus materials
Externí odkaz:
http://arxiv.org/abs/2411.03396
Large language models (LLMs) have demonstrated a remarkable ability to serve as general-purpose tools for various language-based tasks. Recent works have demonstrated that the efficacy of such models can be improved through iterative dialog between m
Externí odkaz:
http://arxiv.org/abs/2411.00053
Autor:
That, Khai Ton, Xu, Chang, Ioannidis, Ioannis, Schneider, Lucas, Posske, Thore, Wiesendanger, Roland, Morr, Dirk K., Wiebe, Jens
Probing spatially confined quantum states from afar - a long-sought goal to minimize external interference - has been proposed to be achievable in condensed matter systems via coherent projection. The latter can be tailored by sculpturing the eigenst
Externí odkaz:
http://arxiv.org/abs/2410.16054
We introduce MDSGen, a novel framework for vision-guided open-domain sound generation optimized for model parameter size, memory consumption, and inference speed. This framework incorporates two key innovations: (1) a redundant video feature removal
Externí odkaz:
http://arxiv.org/abs/2410.02130
Autor:
Ton, Khiem, Nguyen, Nhi, Nazzal, Mahmoud, Khreishah, Abdallah, Borcea, Cristian, Phan, NhatHai, Jin, Ruoming, Khalil, Issa, Shen, Yelong
This paper introduces SGCode, a flexible prompt-optimizing system to generate secure code with large language models (LLMs). SGCode integrates recent prompt-optimization approaches with LLMs in a unified system accessible through front-end and back-e
Externí odkaz:
http://arxiv.org/abs/2409.07368
Publikováno v:
Ecological Informatics, Volume 83, 2024, 102805, ISSN 1574-9541, (https://www.sciencedirect.com/science/article/pii/S1574954124003479)
Camera trap imagery has become an invaluable asset in contemporary wildlife surveillance, enabling researchers to observe and investigate the behaviors of wild animals. While existing methods rely solely on image data for classification, this may not
Externí odkaz:
http://arxiv.org/abs/2409.04825
Open-vocabulary 3D instance segmentation transcends traditional closed-vocabulary methods by enabling the identification of both previously seen and unseen objects in real-world scenarios. It leverages a dual-modality approach, utilizing both 3D poin
Externí odkaz:
http://arxiv.org/abs/2408.08591
This research compares large language model (LLM) fine-tuning methods, including Quantized Low Rank Adapter (QLoRA), Retrieval Augmented fine-tuning (RAFT), and Reinforcement Learning from Human Feedback (RLHF), and additionally compared LLM evaluati
Externí odkaz:
http://arxiv.org/abs/2408.03562