Zobrazeno 1 - 9
of 9
pro vyhledávání: '"Song, Jongyoon"'
Multi-hop reasoning, which requires multi-step reasoning based on the supporting documents within a given context, remains challenging for large language models (LLMs). LLMs often struggle to filter out irrelevant documents within the context, and th
Externí odkaz:
http://arxiv.org/abs/2410.07103
Autor:
Yu, Sangwon, Song, Jongyoon, Hwang, Bongkyu, Kang, Hoyoung, Cho, Sooah, Choi, Junhwa, Joe, Seongho, Lee, Taehee, Gwon, Youngjune L., Yoon, Sungroh
A binary decision task, like yes-no questions or answer verification, reflects a significant real-world scenario such as where users look for confirmation about the correctness of their decisions on specific issues. In this work, we observe that lang
Externí odkaz:
http://arxiv.org/abs/2408.00137
In this paper, we identify a new category of bias that induces input-conflicting hallucinations, where large language models (LLMs) generate responses inconsistent with the content of the input context. This issue we have termed the false negative pr
Externí odkaz:
http://arxiv.org/abs/2406.13929
Autor:
Song, Jongyoon, Park, Nohil, Hwang, Bongkyu, Yun, Jaewoong, Joe, Seongho, Gwon, Youngjune L., Yoon, Sungroh
Abstractive summarization models often generate factually inconsistent content particularly when the parametric knowledge of the model conflicts with the knowledge in the input document. In this paper, we analyze the robustness of fine-tuning based s
Externí odkaz:
http://arxiv.org/abs/2402.15162
Non-autoregressive neural machine translation (NART) models suffer from the multi-modality problem which causes translation inconsistency such as token repetition. Most recent approaches have attempted to solve this problem by implicitly modeling dep
Externí odkaz:
http://arxiv.org/abs/2109.06481
Publikováno v:
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL), Dublin, Ireland, May 2022
Recent studies have determined that the learned token embeddings of large-scale neural language models are degenerated to be anisotropic with a narrow-cone shape. This phenomenon, called the representation degeneration problem, facilitates an increas
Externí odkaz:
http://arxiv.org/abs/2109.03127
Most modern text-to-speech architectures use a WaveNet vocoder for synthesizing high-fidelity waveform audio, but there have been limitations, such as high inference time, in its practical application due to its ancestral sampling scheme. The recentl
Externí odkaz:
http://arxiv.org/abs/1811.02155
Autor:
Lee, Jaekoo, Lee, Byunghan, Song, Jongyoon, Yoon, Jaesik, Lee, Yongsik, Lee, Donghun, Yoon, Sungroh
With a new era of cloud and big data, Database Management Systems (DBMSs) have become more crucial in numerous enterprise business applications in all the industries. Accordingly, the importance of their proactive and preventive maintenance has also
Externí odkaz:
http://arxiv.org/abs/1804.05497
Recent works have shown that synthetic parallel data automatically generated by translation models can be effective for various neural machine translation (NMT) issues. In this study, we build NMT systems using only synthetic parallel data. As an eff
Externí odkaz:
http://arxiv.org/abs/1704.00253