亲爱的研友该休息了!由于当前在线用户较少,发布求助请尽量完整地填写文献信息,科研通机器人24小时在线,伴您度过漫漫科研夜!身体可是革命的本钱,早点休息,好梦!

Pocket GPT: A Lightweight Generative AI-Based Real-Time Language Learning Platform for Enhanced Multilingual Proficiency

计算机科学 生成模型 生成语法 计算机体系结构 人工智能 自然语言处理 嵌入式系统
作者
Qiangwei Li,M. Z. Li
出处
期刊:Journal of Circuits, Systems, and Computers [World Scientific]
卷期号:35 (06)
标识
DOI:10.1142/s0218126625504547
摘要

Language learning is important; traditionally, improving language skills mostly depends on the teacher–student interaction. Instructors who are trained with expert language skills can train the students. But in the modern technological era, we are increasingly focused on devices and human–machine interactions every day. This immediate shift demands well-versed language platforms like ChatGPT and Large Language Models (LLM), which are significantly more advanced in language training through NLP and other language processing models. However, accessibility is one of the limitations that we face regarding this HiFi model. These advanced platforms are not portable to all devices. GPT and other language models are resource-constrained and require effective resources to respond in real-time. To address this, we designed a Pocket GPT model—an advanced GPT app that is portable for all devices (like smartphones, tab and laptops). This novel Pocket GPT involves Generative AI (GenAI) techniques, which encourage human-like interactions in real time. The suggested model combines the advantages of LeBERT (a lightweight BERT) based on powerful language understanding and a masked CNN layer for further enhanced processing efficiency and accuracy. Through interactive exercises, contextual understanding and individualized feedback, this software aims to provide real-time adaptive language learning that can handle various user proficiency levels. The model simulation is conducted using a dataset of 120 Thai students who were unfamiliar with the English language at the age of 19–20. Participants were classified into two groups: control and experimental groups. Control group students depend on traditional language learning systems, and experimental group students can interact with the proposed Pocket GPT with their smartphones or tablets. The results highlight the significant improvement in the English proficiency of the experimental group. Feedback from the focused experimental group interview highlights improved engagement, motivation and confidence in language learning. Thus, the Pocket GPT is a fine contribution to the language learning platform, which can train not only English but is also accessible to train, with all language skills with its pre-trained efficiency.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
5秒前
Feng完成签到 ,获得积分10
20秒前
坦率广山发布了新的文献求助10
24秒前
31秒前
45秒前
坦率广山完成签到 ,获得积分10
1分钟前
UPUP0707完成签到,获得积分10
1分钟前
1分钟前
wanci应助风华正茂采纳,获得30
1分钟前
风华正茂完成签到,获得积分10
2分钟前
2分钟前
风华正茂发布了新的文献求助10
2分钟前
blenx完成签到,获得积分10
2分钟前
olekravchenko发布了新的文献求助10
2分钟前
2分钟前
2分钟前
朱羊羊发布了新的文献求助10
2分钟前
3分钟前
3分钟前
3分钟前
佳宝(不可以喝但能吃完成签到,获得积分10
3分钟前
3分钟前
3分钟前
记ds发布了新的文献求助10
3分钟前
卷卷卷儿完成签到 ,获得积分10
3分钟前
科研通AI6应助han采纳,获得10
4分钟前
黑摄会阿Fay完成签到,获得积分10
4分钟前
活泼的鼠标完成签到 ,获得积分10
4分钟前
4分钟前
4分钟前
凸凸完成签到 ,获得积分10
4分钟前
顾矜应助风华正茂采纳,获得10
4分钟前
4分钟前
4分钟前
慕青应助文章多多采纳,获得10
4分钟前
David发布了新的文献求助10
4分钟前
4分钟前
凸凸发布了新的文献求助10
4分钟前
领导范儿应助科研通管家采纳,获得10
5分钟前
英姑应助科研通管家采纳,获得10
5分钟前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
Encyclopedia of Reproduction Third Edition 3000
Comprehensive Methanol Science Production, Applications, and Emerging Technologies 2000
化妆品原料学 1000
Psychology of Self-Regulation 600
1st Edition Sports Rehabilitation and Training Multidisciplinary Perspectives By Richard Moss, Adam Gledhill 600
Red Book: 2024–2027 Report of the Committee on Infectious Diseases 500
热门求助领域 (近24小时)
化学 材料科学 生物 医学 工程类 计算机科学 有机化学 物理 生物化学 纳米技术 复合材料 内科学 化学工程 人工智能 催化作用 遗传学 数学 基因 量子力学 物理化学
热门帖子
关注 科研通微信公众号,转发送积分 5639664
求助须知:如何正确求助?哪些是违规求助? 4749580
关于积分的说明 15007025
捐赠科研通 4797830
什么是DOI,文献DOI怎么找? 2563907
邀请新用户注册赠送积分活动 1522813
关于科研通互助平台的介绍 1482510