Negative / Null Result ReportOpen accessComputer Science
Hoyeon Chang, Jinho Park, Seonghyeon Ye et al. · 2024 · arXiv
Despite the recent observation that large language models (LLMs) can store substantial factual knowledge, there is a limited understanding of the mechanisms of how they acquire factual knowledge through pretraining. This work addresses this gap by studying how LLMs acquire factual knowledge during pretraining. The findings reveal several important insights into the dynamics of factual knowledge acquisition during pretraining. First, counterintuitively, we observe that pretraining on more data shows no significant improvement in the model's capability to acquire and maintain factual knowledge.
Negative / Null Result ReportOpen accessEngineering
Junwei Yang, Xiao-Xin Li, Feihong Liu et al. · 2021 · arXiv
Recent studies on T1-assisted MRI reconstruction for under-sampled images of other modalities have demonstrated the potential of further accelerating MRI acquisition of other modalities. Most of the state-of-the-art approaches have achieved improvement through the development of network architectures for fixed under-sampling patterns, without fully exploiting the complementary information between modalities. Although existing under-sampling pattern learning algorithms can be simply modified to allow the fully-sampled T1-weighted MR image to assist the pattern learning, no significant improveme
Negative / Null Result Report
· 2022 · The Brown University Psychopharmacology Update
Eight infusions of ketamine administered over 4 weeks did not significantly improve symptoms of post‐traumatic stress disorder (PTSD) in a group of veterans and active‐duty military who had failed on prior antidepressant treatment, a…
View details →DOI: 10.1002/pu.30864 Negative / Null Result ReportOpen accessComputer Science
Ivan Silajev, Niels Victor, Phillip Mortimer · 2022 · arXiv
This paper presents an application of the LayoutLMv3 model for semantic table detection on financial documents from the IIIT-AR-13K dataset. The motivation behind this paper's experiment was that LayoutLMv3's official paper had no results for table detection using semantic information. We concluded that our approach did not improve the model's table detection capabilities, for which we can give several possible reasons. Either the model's weights were unsuitable for our purpose, or we needed to invest more time in optimising the model's hyperparameters. It is also possible that semantic inform
Negative / Null Result ReportOpen accessComputer Science
Bob de Ruiter · 2021 · arXiv
The goal of this study was to improve the post-processing of precipitation forecasts using convolutional neural networks (CNNs). Instead of post-processing forecasts on a per-pixel basis, as is usually done when employing machine learning in meteorological post-processing, input forecast images were combined and transformed into probabilistic output forecast images using fully convolutional neural networks. CNNs did not outperform regularized logistic regression. Additionally, an ablation analysis was performed. Combining input forecasts from a global low-resolution weather model and a regiona
Negative / Null Result ReportOpen accessComputer Science
Haresh Rengaraj Rajamohan, Yuxuan Chen, Kyunghyun Cho et al. · 2026 · arXiv
This study assesses whether self-supervised learning (SSL) improves knee osteoarthritis (OA) modeling for diagnosis and prognosis relative to ImageNet-pretrained initialization. We compared (i) image-only SSL pretrained on knee radiographs from the OAI, MOST, and NYU cohorts, and (ii) multimodal image-text SSL pretrained on hospital knee radiographs paired with radiologist impressions. For diagnostic Kellgren-Lawrence (KL) grade prediction, SSL yielded mixed results. While image-only SSL improved accuracy during linear probing (frozen encoder), it did not outperform ImageNet pretraining during
Failed Experiment Report
Mrityunjoy Hazra, Ashok Kumar Singh · 2023 · International Journal of Engineering Materials and Manufacture
Present work describes the metallurgical failure investigation of a failed towing arm/bar. Aircraft attached with this incident travelled a distance of 2 km with normal towing speed (8 km/hr). The component is said to have broken while…
View details →DOI: 10.26776/ijemm.08.03.2023.01 Negative / Null Result ReportOpen accessComputer Science
N. J. Ayres · 2018 · arXiv
We report on a search for ultra-low-mass axion-like dark matter by analysing the ratio of the spin-precession frequencies of stored ultracold neutrons and $^{199}$Hg atoms for an axion-induced oscillating electric dipole moment of the neutron and an axion-wind spin-precession effect. No signal consistent with dark matter is observed for the axion mass range $10^{-24}~\textrm{eV} \le m_a \le 10^{-17}~\textrm{eV}$. Our null result sets the first laboratory constraints on the coupling of axion dark matter to gluons, which improve on astrophysical limits by up to 3 orders of magnitude, and also im
Negative / Null Result ReportOpen accessPhysics
Michael Blomqvist, Edvard Mortsell, Serena Nobili · 2008 · arXiv
We discuss the possibility to identify anisotropic and/or inhomogeneous cosmological models using type Ia supernova data. A search for correlations in current type Ia peak magnitudes over a large range of angular scales yields a null result. However, the same analysis limited to supernovae at low redshift, shows a feeble anticorrelation at the two sigma level at angular scales of about 40 degrees. Upcoming data from, e.g., the SNLS (Supernova Legacy Survey) and the SDSS-II (SDSS: Sloan Digital Sky Survey) supernova searches will improve our limits on the size of - or possibly detect - possible
Negative / Null Result Report
· 2019 · The Brown University Psychopharmacology Update
Augmenting sertraline with eicosapentaenoic acid (EPA) for 10 weeks did not improve depressive symptoms relative to placebo in a group of 144 patients with major depressive disorder and existing or high risk of heart disease, a randomized…
View details →DOI: 10.1002/pu.30472 Negative / Null Result Report
· 2019 · The Brown University Psychopharmacology Update
Six weeks of D‐cycloserine treatment for patients with depression who had responded to doses of ketamine did not improve depressive symptoms compared with placebo, a study has found. D‐cycloserine was found to maintain the anti‐suicidal…
View details →DOI: 10.1002/pu.30506 Negative / Null Result ReportOpen accessComputer Science
C. Abel, N. J. Ayres, G. Ban et al. · 2017 · arXiv
We report on a search for ultra-low-mass axion-like dark matter by analysing the ratio of the spin-precession frequencies of stored ultracold neutrons and $^{199}$Hg atoms for an axion-induced oscillating electric dipole moment of the neutron and an axion-wind spin-precession effect. No signal consistent with dark matter is observed for the axion mass range $10^{-24}~\textrm{eV} \le m_a \le 10^{-17}~\textrm{eV}$. Our null result sets the first laboratory constraints on the coupling of axion dark matter to gluons, which improve on astrophysical limits by up to 3 orders of magnitude, and also im
Negative / Null Result ReportOpen accessEconomics, Econometrics and Finance
Ruiwu Liu · 2021 · arXiv
Acemoglu and Johnson (2007) put forward the unprecedented view that health improvement has no significant effect on income growth. To arrive at this conclusion, they constructed predicted mortality as an instrumental variable based on the WHO international disease interventions to analyse this problem. I replicate the process of their research and eliminate some biases in their estimate. In addition, and more importantly, we argue that the construction of their instrumental variable contains a violation of the exclusion restriction of their instrumental variable. This negative correlation betw
Negative / Null Result Report
Cui L, Harrison L, Liu J · 2022 · Preprint
Many studies have failed to replicate the disfluency effect (i.e. disfluent font better than fluent font) on Cognitive Reflection Test (CRT) performance shown in Alter et al. (2007). Those studies manipulated perceptual disfluency using…
View details →DOI: 10.31234/osf.io/qdxw3 