Zobrazeno 1 - 10
of 72
pro vyhledávání: '"Bae Sangmin"'
Large language models (LLMs) are expensive to deploy. Parameter sharing offers a possible path towards reducing their size and cost, but its effectiveness in modern LLMs remains fairly limited. In this work, we revisit "layer tying" as form of parame
Externí odkaz:
http://arxiv.org/abs/2410.20672
Autor:
Yang, Yongjin, Kim, Sihyeon, Jung, Hojung, Bae, Sangmin, Kim, SangMook, Yun, Se-Young, Lee, Kimin
Fine-tuning text-to-image diffusion models with human feedback is an effective method for aligning model behavior with human intentions. However, this alignment process often suffers from slow convergence due to the large size and noise present in hu
Externí odkaz:
http://arxiv.org/abs/2410.10166
Despite the astonishing performance of recent Large Vision-Language Models (LVLMs), these models often generate inaccurate responses. To address this issue, previous studies have focused on mitigating hallucinations by employing contrastive decoding
Externí odkaz:
http://arxiv.org/abs/2408.05337
Autor:
Choi, Yunseon, Bae, Sangmin, Ban, Seonghyun, Jeong, Minchan, Zhang, Chuheng, Song, Lei, Zhao, Li, Bian, Jiang, Kim, Kee-Eung
With the advent of foundation models, prompt tuning has positioned itself as an important technique for directing model behaviors and eliciting desired responses. Prompt tuning regards selecting appropriate keywords included into the input, thereby a
Externí odkaz:
http://arxiv.org/abs/2407.14733
Audio-visual speech recognition (AVSR) aims to transcribe human speech using both audio and video modalities. In practical environments with noise-corrupted audio, the role of video information becomes crucial. However, prior works have primarily foc
Externí odkaz:
http://arxiv.org/abs/2407.03563
Autor:
Ho, Namgyu, Bae, Sangmin, Kim, Taehyeon, Jo, Hyunjik, Kim, Yireun, Schuster, Tal, Fisch, Adam, Thorne, James, Yun, Se-Young
We introduce the Block Transformer which adopts hierarchical global-to-local modeling to autoregressive transformers to mitigate the inference bottlenecks associated with self-attention. Self-attention requires the key-value (KV) cache of all previou
Externí odkaz:
http://arxiv.org/abs/2406.02657
The recently introduced TabPFN pretrains an In-Context Learning (ICL) transformer on synthetic data to perform tabular data classification. As synthetic data does not share features or labels with real-world data, the underlying mechanism that contri
Externí odkaz:
http://arxiv.org/abs/2405.13396
Recent advancements in AI have democratized its deployment as a healthcare assistant. While pretrained models from large-scale visual and audio datasets have demonstrably generalized to this task, surprisingly, no studies have explored pretrained spe
Externí odkaz:
http://arxiv.org/abs/2405.02996
Despite the remarkable advances in deep learning technology, achieving satisfactory performance in lung sound classification remains a challenge due to the scarcity of available data. Moreover, the respiratory sound samples are collected from a varie
Externí odkaz:
http://arxiv.org/abs/2312.09603
Autor:
Kim, Yujin, Yoon, Jaehong, Ye, Seonghyeon, Bae, Sangmin, Ho, Namgyu, Hwang, Sung Ju, Yun, Se-young
The dynamic nature of knowledge in an ever-changing world presents challenges for language models trained on static data; the model in the real world often requires not only acquiring new knowledge but also overwriting outdated information into updat
Externí odkaz:
http://arxiv.org/abs/2311.08106