Efficient large language model with analog in-memory computing.

Nat Comput Sci

Department of Computer Science, Royal Holloway, University of London, Egham, UK.

Published: January 2025

Download full-text PDF

Source
http://dx.doi.org/10.1038/s43588-024-00760-yDOI Listing

Publication Analysis

Top Keywords

efficient large
4
large language
4
language model
4
model analog
4
analog in-memory
4
in-memory computing
4
efficient
1
language
1
model
1
analog
1

Similar Publications

Efficient production system for hydrogel-based transparent soil for plant root observation.

Biotechniques

January 2025

Department of Applied Biological Chemistry, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Tokyo, Japan.

Observation of plant root morphology in soil is of fundamental importance in plant research, but the lack of transparency of the soil hampers direct observation of roots. One of the approaches to overcome this technical limitation is the use of "transparent soil" (TS), hydrogel-based beads produced by spherification of gelling agents. However, the production of TS by natural dripping of gelling solution can be labor intensive, time consuming and difficult to maintain consistent product quality.

View Article and Find Full Text PDF

Carbonic anhydrases (CAs) are ubiquitous enzymes that catalyze reversibly both the hydration and dehydration reactions of CO and HCO-, respectively. Higher plants contain many different isoforms of CAs that can be classified into α-, β- and γ-type subfamilies. β-type CAs play a key role in the CO-concentrating mechanism, thereby contributing to efficient photosynthesis in the C plants in addition to many other biochemical reactions in plant metabolism.

View Article and Find Full Text PDF

The influence of high-intensity electric fields on the stability of polymeric materials is a problem of interest in the design of next-generation energy storage and electronic devices, and for understanding the limits of stability of polymer films exposed to large electric fields generally. Here, we show that the dielectric strength of entangled glassy polymer films increases as an inverse power-law of the film thickness for "ultrathin" films below a micron in thickness. The dielectric strength enhancement in these polymer films becomes as large as ≈2 GV/m in films thinner than 100 nm, but in this thickness regime, the increase of the dielectric strength depends strongly on the polymer mass, sample aging time, and the method of film preparation.

View Article and Find Full Text PDF

A comparison of the persuasiveness of human and ChatGPT generated pro-vaccine messages for HPV.

Front Public Health

January 2025

CAS Key Laboratory of Behavioral Science, Institute of Psychology, Chinese Academy of Sciences, Beijing, China.

Introduction: Public health messaging is crucial for promoting beneficial health outcomes, and the latest advancements in artificial intelligence offer new opportunities in this field. This study aimed to evaluate the effectiveness of ChatGPT-4 in generating pro-vaccine messages on different topics for Human Papillomavirus (HPV) vaccination.

Methods: In this study ( = 60), we examined the persuasive effect of pro-vaccine messages generated by GPT-4 and humans, which were constructed based on 17 factors impacting HPV vaccination.

View Article and Find Full Text PDF

The solute carrier (SLC) family of membrane proteins is a large class of transporters for many small molecules that are vital for cellular function. Several pathogenic mutations are reported in the glucose transporter subfamily SLC2, causing Glut1-deficiency syndrome (GLUT1DS1, GLUT1DS2), epilepsy (EIG2) and cryohydrocytosis with neurological defects (Dystonia-9). Understanding the link between these mutations and transporter dynamics is crucial to elucidate their role in the dysfunction of the underlying transport mechanism, which we investigate using molecular dynamics simulations.

View Article and Find Full Text PDF

Want AI Summaries of new PubMed Abstracts delivered to your In-box?

Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!