Search results
Sep 20, 2024 · A trailblazing fluid mechanics project led by City College of New York Grove School of Engineering Professor Taehun Lee is one of 16 initiatives nationally awarded supercomputing time at the Argonne Leadership Computing Facility (ALCF) by the U.S. Department of Energy’s (DoE) Advanced Scientific Computing Research (ASCR) Leadership Computing Challenge (ALCC).
4 days ago · In this work, we show the pre-trained language models return distinguishable generation probability and uncertainty distribution to unfaithfully hallucinated texts, regardless of their size and structure. By examining 24 models on 6 data sets, we find out that 88-98% of cases return statistically significantly distinguishable generation probability and uncertainty distributions. Using this general phenomenon, we showcase a hallucination-reducing training algorithm. Our algorithm outperforms ...
4 days ago · Pre-trained Language Models Return Distinguishable Probability Distributions to Unfaithfully Hallucinated Texts: Paper and Code. In this work, we show the pre-trained language models return distinguishable generation probability and uncertainty distribution to unfaithfully hallucinated texts, regardless of their size and structure. By examining 24 models on 6 data sets, we find out that 88-98% of cases return statistically significantly distinguishable generation probability and uncertainty ...
4 days ago · 25 Sep 2024 · Taehun Cha, Donghun Lee · Edit social preview In this work, we show the pre-trained language models return distinguishable generation probability and uncertainty distribution to unfaithfully hallucinated texts, regardless of their size and structure.
Sep 12, 2024 · the Altmetric Attention Score and how the score is calculated. The relentless miniaturization inherent in complementary metal-oxide semiconductor technology has created challenges at the interface of two-dimensional (2D) materials and metal electrodes. These c...
Sep 19, 2024 · The proposed ICGNet firstly utilizes a reverse-contour guidance module to aggregate low-level edge detail information and meanwhile constraint reverse region and an innovative hybrid pyramid pooling fusion module fuses the multi-level features generated from the decoder in the case of considering salient features and less background.
1 day ago · Jared Redmond, piano | Taehun Lee, guitar free improvisation performance. Angelism Thurs., Dec. 29th, 2022 9pm Jeungsan-dong, Seoul, KR
