13197 research outputs found
Sort by
Can we obtain significant success in RST discourse parsing by using Large Language Models?
No full textRecently, decoder-only pre-trained large language models (LLMs), with several tens of billion parameters, have significantly impacted a wide range of natural language processing (NLP) tasks. While encoder-only or encoder-decoder pre-trained language models have already proved to be effective in discourse parsing, the extent to which LLMs can perform this task remains an open research question. Therefore, this paper explores how beneficial such LLMs are for Rhetorical Structure Theory (RST) discourse parsing. Here, the parsing process for both fundamental top-down and bottom-up strategies is converted into prompts, which LLMs can work with. We employ Llama 2 and fine-tune it with QLoRA, which has fewer parameters that can be tuned. Experimental results on three benchmark datasets, RST-DT, Instr-DT, and the GUM corpus, demonstrate that Llama 2 with 70 billion parameters in the bottom-up strategy obtained state-of-the-art (SOTA) results with significant differences. Furthermore, our parsers demonstrated generalizability when evaluated on RST-DT, showing that, in spite of being trained with the GUM corpus, it obtained similar performances to those of existing parsers trained with RST-DT.conference pape
Surface zeta potential and protein adsorption on the coating surface of a heteroarm star polymer with a controlled hydrophilic/hydrophobic arm ratio
Get PDFA surface coated with a star polymer is believed to form a highly dense polymer brush-like architecture and inhibit biofouling. In this study, the surface properties of the star polymer coating were evaluated with their resistance to protein adsorption and surface zeta (ζ)-potential to clarify the mechanism for inhibition of cell adhesion. The surface of the star polymer coating with a high density of poly(2-hydroxyethyl methacrylate) (PHEMA) formed an electrically neutral diffuse brush structure in water and showed high resistance to protein adsorption. Considering the data obtained in the study, the surface ζ-potential and antibiofouling properties were correlated by controlling the molecular architecture of the coating material.journal articl
Comb Polyurethanes Consisting of Hard Segment Backbones and Dangling Soft Segments for Tailoring Mechanical Properties of Thermoplastics
Get PDFGraft polymers, including comb and bottlebrush polymers, have emerged as a topological molecular design for tailoring elastic material properties, relying on structural parameters. Yet, in stiffer materials, particularly thermoplastics, the inherent chemical nature of polymers often eclipses this topological design of mechanical properties. Here, we present comb polyurethanes (PUs) with hard segments in the backbone and soft segments in the side chains, with varying structural parameters, side chain spacing, and length. A series of comb PUs were synthesized by combining 4,4′-methylenebis(cyclohexylisocyanate), ethylene glycol, and two types of α-methyl, ω-diol-terminated polyether macromonomers in a polyaddition procedure. These macromonomers with a controlled average molecular mass were synthesized through end group modification of methyl-terminated polyethylene glycol and cationic ring-opening polymerization of THF. Interestingly, the comb PUs displayed a linear correlation between various mechanical properties and backbone volume fraction (φHS), including Young’s modulus, tensile strength, yielding strength, and elongation at break. FT-IR and rheological measurements indicated that the backbones approach each other more freely than scaling analysis predicted. We hypothesized that comb PU has a network structure composed of hard segments diluted by viscoelastic soft segment side chains. Overall, this work highlights the unique mechanical properties of the comb PUs with hydrogen bonds in the backbone in topological designs for thermoplastic mechanical properties.journal articl
Understanding Newcomer Activities Prior to Onboarding Open Source Software (OSS) Projects on GitHub
Get PDF奈良先端科学技術大学院大学博士(工学)doctoral thesi
The extracellular vesicles secreted from IRSp53-mediated membrane protrusions promote cancer cell proliferation via integrin and laminin
Get PDF奈良先端科学技術大学院大学博士(バイオサイエンス)doctoral thesi
ツリ ジョウタツ シエン システム ノ ジツゲン ニ ムケタ リール ソウチャク センサ オ モチイタ キャスト シュルイ ノ ブンルイ オヨビ キャスト ノ ジュクレンド ハンテイ ニ カンスル ケンキュウ
Get PDF奈良先端科学技術大学院大学修士(工学)master thesi
