The explosive growth of lithium-ion battery literature has led to severe knowledge overload,challenging researchers'ability to efficiently extract structured information.While large language models(LLMs)offer cons...The explosive growth of lithium-ion battery literature has led to severe knowledge overload,challenging researchers'ability to efficiently extract structured information.While large language models(LLMs)offer considerable potential for automating this task,their practical application in scientific domains is nonetheless constrained by high application programming interface(API)costs and computational resources required for fine-tuning.To address these limitations,a cognition-enhanced instruction framework(CEIF)is proposed,wherein a high-performance teacher model(such as DeepSeek-R1)provides dynamic feedback,prompt refinement,and training data optimization to guide the learning process of low-parameter models.Experimental results demonstrate that the low-parameter models(6B-9B)optimized via the CEIF achieve approximately 85%accuracy in battery literature extraction tasks,approaching the performance of GPT-4 while requiring only a single NVIDIA RTX 3090 GPU.Furthermore,the emergence of an"Aha moment"characterized by rapid performance improvement during specialized learning is observed,offering novel theoretical insights for the design and optimization of domainspecific models.展开更多
Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interact...Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interactions to predict future items of interest.However,many current methods rely on unique user and item IDs,limiting their ability to represent users and items effectively,especially in zero-shot learning scenarios where training data is scarce.With the rapid development of Large Language Models(LLMs),researchers are exploring their potential to enhance recommendation systems.However,there is a semantic gap between the linguistic semantics of LLMs and the collaborative semantics of recommendation systems,where items are typically indexed by IDs.Moreover,most research focuses on item representations,neglecting personalized user modeling.To address these issues,we propose a sequential recommendation framework using LLMs,called CIT-Rec,a model that integrates Collaborative semantics for user representation and Image and Text information for item representation to enhance Recommendations.Specifically,by aligning intuitive image information with text containing semantic features,we can more accurately represent items,improving item representation quality.We focus not only on item representations but also on user representations.To more precisely capture users’personalized preferences,we use traditional sequential recommendation models to train on users’historical interaction data,effectively capturing behavioral patterns.Finally,by combining LLMs and traditional sequential recommendation models,we allow the LLM to understand linguistic semantics while capturing collaborative semantics.Extensive evaluations on real-world datasets show that our model outperforms baseline methods,effectively combining user interaction history with item visual and textual modalities to provide personalized recommendations.展开更多
Objective To investigate methods for constructing a high-quality instructional dataset for traditional Chinese medicine(TCM)mental disorders and to validate its efficacy.Methods We proposed the Fine-Med-Mental-T&P...Objective To investigate methods for constructing a high-quality instructional dataset for traditional Chinese medicine(TCM)mental disorders and to validate its efficacy.Methods We proposed the Fine-Med-Mental-T&P methodology for constructing high-quality instruction datasets in TCM mental disorders.This approach integrates theoretical knowledge and practical case studies through a dual-track strategy.(i)Theoretical track:textbooks and guidelines on TCM mental disorders were manually segmented.Initial responses were generated using DeepSeek-V3,followed by refinement by the Qwen3-32B model to align the expression with human preferences.A screening algorithm was then applied to select 16000 high-quality instruction pairs.(ii)Practical track:starting from over 600 real clinical case seeds,diagnostic and therapeutic instruction pairs were generated using DeepSeek-V3 and subsequently screened through manual evaluation,resulting in 4000 high-quality practiceoriented instruction pairs.The integration of both tracks yielded the Med-Mental-Instruct-T&P dataset,comprising a total of 20000 instruction pairs.To validate the dataset’s effectiveness,three experimental evaluations(both manual and automated)were conducted:(i)comparative studies to compare the performance of models fine-tuned on different datasets;(ii)benchmarking to compare against mainstream TCM-specific large language models(LLMs);(iii)data ablation study to investigate the relationship between data volume and model performance.Results Experimental results demonstrate the superior performance of T&P-model finetuned on the Med-Mental-Instruct-T&P dataset.In the comparative study,the T&P-model significantly outperformed the baseline models trained solely on self-generated or purely human-curated baseline data.This superiority was evident in both automated metrics(ROUGEL>0.55)and expert manual evaluations(scoring above 7/10 across accuracy).In benchmark comparisons,the T&P-model also excelled against existing mainstream TCM LLMs(e.g.,HuatuoGPT and ZuoyiGPT).It showed particularly strong capabilities in handling diverse clinical presentations,including challenging disorders such as insomnia and coma,showcasing its robustness and versatility.Data ablation studies showed that T&P-model performance had an overall upward trend with minor fluctuations when training data increased from 10%to 50%;beyond 50%,performance improvement slowed significantly,with metrics plateauing and approaching a saturation point.展开更多
This paper firstly discusses five models of bilingual education experimented or applied in different countries of the world.Then it puts forward two feasible models of classroom bilingual instruction—the high level a...This paper firstly discusses five models of bilingual education experimented or applied in different countries of the world.Then it puts forward two feasible models of classroom bilingual instruction—the high level and the low level.Finally it tentatively points out the course criteria of classroom bilingual instruction in Chinese universities,which are believed to be of considerable reference value to bilingual instruction in China.展开更多
SPSS statistical analysis system is used to examine the teaching effects of three different instruction models which are commonly used in military English teaching. Of the three models, interactive teaching model has ...SPSS statistical analysis system is used to examine the teaching effects of three different instruction models which are commonly used in military English teaching. Of the three models, interactive teaching model has shown to have advantage over the other two, the situational teaching model and the task-based Instruction model. It is concluded that it is beneficial to improve teaching effects in military English teaching if the teacher can stimulate students to involve themselves more in class. To do so, it is necessary to strike a balance between the learners＇ language input and output so as to ensure that the input knowledge be consolidated and digested through practice in output.展开更多
The paper presents a case study on listening-speaking class instruction models based on exploratory practice. Assuming the dual roles of a teacher and a researcher, the writer of this paper has implemented three model...The paper presents a case study on listening-speaking class instruction models based on exploratory practice. Assuming the dual roles of a teacher and a researcher, the writer of this paper has implemented three models of classroom instructions for the Viewing, Listening ＆ Speaking Class. From the students＇ report, teacher＇s log and classroom observation, it is concluded that the three models meet the needs of different students. The key to a successful Viewing, Listening＆ Speaking Class is to set reasonable goals and analyze students＇ needs.展开更多
Online synchronous educational experience via videoconferencing has become ubiquitous domestically and internationally with the spread of Covid-19 Virus.Online teaching theories and educational experiences through com...Online synchronous educational experience via videoconferencing has become ubiquitous domestically and internationally with the spread of Covid-19 Virus.Online teaching theories and educational experiences through computer-mediated communication have been integrating and evolving during the current situation.The objective of this paper is to demonstrate a practical teaching model for Chinese college students’English as a foreign language learning course.The key finding of the research is that the 5-step-teaching Model developed under the context of online synchronous learning and teaching is both functional and efficient on the theoretical basis of constructivist approach.This teaching model is proved to be practical,interesting,useful and functional to both second-years and graduates during the implementation of educational transaction according to the survey conducted at the end of the semester.The ubiquity of online synchronous learning and teaching would lead to a meaningful and worthwhile educational experience and have a far-reaching implication for both educators and student’s personal development,especially students’critical thinking capability.The findings of the paper are pedagogically valuable for e-teachers in conducting online courses.It also holds true that the management of education and society could benefit and adopt this teaching model to various courses in different institutions.This study is hopefully to provide research-and-experience-based findings to enrich course design and instructional approaches and educational experiences.Thus,it is of interest to online teachers,instructional designers and programme developers in the field of education.展开更多
Generative artificial intelligence,represented by large language models,holds vast application scenarios and significant development potential in the field of language teaching.This study employs large language models...Generative artificial intelligence,represented by large language models,holds vast application scenarios and significant development potential in the field of language teaching.This study employs large language models such as ChatGPT4o,ERNIE Bot,and Spark Cognition to explore how they empower teachers in international Chinese language teaching through practical cases.It focuses on various aspects of international Chinese language teaching and language skills training,examining the application effects of large language models in generating tailored teaching content and converting textual content into multimodal teaching materials.Finally,the study proposes that teachers should rationally recognize the opportunities and challenges that large language models bring to the teaching ecosystem,while acknowledging the models’efficiency in empowering teachers’instruction,it is crucial to fully recognize their essential tool nature,uphold teachers’subjectivity,and pay close attention to the boundaries of their development and application.展开更多
We have come to know comprehension better by studying the nature and characteristics of reading processes and by weighing the pros and cons of different reading theories and their influence on the teaching of reading....We have come to know comprehension better by studying the nature and characteristics of reading processes and by weighing the pros and cons of different reading theories and their influence on the teaching of reading. But in China, unfortunately, few achievements have been made in the field of reading theory research. Thus, in order to ameliorate our reading teaching, it is urgent for us to set up our own system of reading theories. The author, after studying carefully the present reading models in China and by absorbing the quintessence of foreign reading theories, ventures to suggest the “discourse word discourse” model. The model gives full thought to the characteristics of Chinese college students and thus suits a reading classroom in China better. However, the comprehension factors involved in the model are far from exhaustive and the model itself leaves much room for modification and addition.\;展开更多
Guided by the"Healthy China 2030"strategy,improving national nutrition and health literacy has become a core task in public health system development.The National Nutrition Plan(2017-2030)explicitly calls fo...Guided by the"Healthy China 2030"strategy,improving national nutrition and health literacy has become a core task in public health system development.The National Nutrition Plan(2017-2030)explicitly calls for"strengthening the training of nutrition talents"and"promoting nutrition science education".As a key vehicle for this mission,the Food Nutrition and Health course in higher education urgently needs to address bottlenecks in traditional teaching,such as low knowledge application and transfer rates,insufficient student engagement,and ineffective guidance on healthy behaviors.The BOPPPS teaching model,with its structured design(Bridge-in,Objective,Pre-assessment,Participatory Learning,Post-assessment,Summary),effectively promotes the internalization of nutritional knowledge and the transformation into healthy behaviors among students by emphasizing practice-oriented teaching activities.In this study,focusing on this course,an in-depth exploration of curriculum teaching design was conducted based on the BOPPPS instructional model,aiming to deeply integrate the strategic objectives of Healthy China into the curriculum,and promote the transformation of nutritional knowledge into healthy decision-making ability.This study provides new insights for food and nutrition education.展开更多
From 2010,a drastic increase in international students’enrollment in many higher education institutions in China has led to the growth of English-medium instruction(EMI)practices and relevant research.With many state...From 2010,a drastic increase in international students’enrollment in many higher education institutions in China has led to the growth of English-medium instruction(EMI)practices and relevant research.With many state-level and province-level initiatives to promote the design and development of model EMI courses,international students’perspectives of these courses are not adequately reflected in the international literature which is primarily done in European higher education contexts.This study takes a qualitative approach by analyzing international students’EMI courses reflection reports,focus group interviews and observation notes of video-recorded EMI courses to probe into their perspectives on EMI practices in a business institute in a university of Shanghai.The findings show that international students show concerns about the systematic design of the curriculum,pedagogical problems and contextual limitations in EMI practices.All EMI stakeholders like curriculum developers,teaching administrators,teachers and students are responsible for these problems.Therefore,meaningful and consistent communication should be greatly promoted among all EMI practices stakeholders.The article concludes with recommendations for further research on model EMI courses in China.展开更多
As we have entered 21st century,new technology has had a great impact on educational instructions. In this paper,I will demonstrate how new technology influences the educational instructions. Then,I try to construct a...As we have entered 21st century,new technology has had a great impact on educational instructions. In this paper,I will demonstrate how new technology influences the educational instructions. Then,I try to construct a new instruction model——"classroom+internet"model. I will explain the characteristics of new technology and its impact on education. The final part is summary.展开更多
The study describes the improvements of the participants in ESP course incorporating Sheltered Instruction Observation Protocol(SIOP) instruction in China. It aims to investigate whether or not it is effective to inco...The study describes the improvements of the participants in ESP course incorporating Sheltered Instruction Observation Protocol(SIOP) instruction in China. It aims to investigate whether or not it is effective to incorporate the sheltered instruction into ESP course and whether or not the participants will be motivated to improve their English skills for professional purposes. Quantitative and qualitative data sources were employed, including English language tests, questionnaires, and interviews.Findings reveal that the sheltered instruction was helpful to ESP course and produced significant achievements in participants' English vocabulary and skills. The most interesting result is that the participants' confidence and interest in learning professional English for future jobs has been improved much.展开更多
Instruction cues are widely employed for research on neural mechanisms during movement preparation.However,their influence on brain connectivity during movement has not received much attention.Herein,15 healthy subjec...Instruction cues are widely employed for research on neural mechanisms during movement preparation.However,their influence on brain connectivity during movement has not received much attention.Herein,15 healthy subjects completed two experimental tasks including either instructed or voluntary movements;meanwhile electroencephalogram(EEG)data were synchronously recorded.Based on source analysis and related literature,six movement-related brain regions were selected,including the left/right supplementary motor area(SMA),left/right inferior frontal gyrus(iFg),and left/right postcentral gyrus(pCg).After assuming 10 a priori models of regional brain connectivity,we evaluated the optimal connectivity model between brain regions for the two scenarios using the dynamic causality model(DCM).During voluntary movement,the movement originated in the SMA,passed through the iFg of the prefrontal lobe,and then returned to the main sensory cortex of the pCg.In the instructed movement,the movement originated in the iFg,and then was transmitted to the pCg and the SMA,as well as from the pCg to the SMA.In contrast to the preparation process of voluntary movement,there were long-range information interactions between the iFg and pCg.Further,almost the same brain regions were active during movement preparation under both voluntary and instructed movement tasks,which evidences certain similarities in dynamic brain connectivity,that is,the brain has direct connections between the bilateral SMA,bilateral pCg,and bilateral SMA,indicating that the both brain hemispheres work together during the movement preparation phase.The results suggest that the network during the preparation process of instructed movements is more complex than voluntary movements.展开更多
Instruction fine-tuning is a key method for adapting large language models(LLMs)to domain-specific tasks,and instruction quality significantly impacts model performance after fine-tuning.Hence,evaluating the quality o...Instruction fine-tuning is a key method for adapting large language models(LLMs)to domain-specific tasks,and instruction quality significantly impacts model performance after fine-tuning.Hence,evaluating the quality of instruction and selecting high-quality instructions are essential steps in the process of LLM instruction fine-tuning.Although existing studies provide important theoretical foundations and techniques for this,there is still room for improvement in terms of generality,the relationship between methods and experimental verification.Current methods for evaluating instruction quality can be classified into four main categories:human evaluation,statistics-based evaluation,model-based evaluation,and LLMs-based evaluation.Among these methods,human evaluation relies on the subjective judgment and domain expertise of the evaluators,which offers interpretability and is suitable for scenarios involving small-scale data and sufficient budgets.Statistics-based evaluation estimates the quality of instructions using indicators such as stopwords and lexical diversity,providing high efficiency and a suitable evaluation for large-scale data.Model-based evaluation employs specific models to quantify indicators such as perplexity(PPL)and instruction following difficulty(IFD),which is flexible and suitable for specific tasks.The LLMs-based evaluation rates the quality of instructions through prompt-based interaction with LLMs,focusing on aspects such as accuracy and coherence,which is highly automated and customizable,simplifying the evaluation process.Finally,considering the limitations of current quality evaluation methods,some future research directions are proposed for improvement.These include refining instruction categories,extending evaluation indicators,enhancing human-AI interaction evaluation method,applying agents in instruction quality evaluation,and developing a comprehensive evaluation framework.展开更多
The development of large language models(LLMs)has created transformative opportunities for the financial industry,especially in the area of financial trading.However,how to integrate LLMs with trading systems has beco...The development of large language models(LLMs)has created transformative opportunities for the financial industry,especially in the area of financial trading.However,how to integrate LLMs with trading systems has become a challenge.To address this problem,we propose an intelligent trade order recognition pipeline that enables the conversion of trade orders into a standard format for trade execution.The system improves the ability of human traders to interact with trading platforms while addressing the problem of misinformation acquisition in trade execution.In addition,we create a trade order dataset of 500 pieces of data to simulate the real-world trading scenarios.Moreover,we design several metrics to provide a comprehensive assessment of dataset reliability and the generative power of big models in finance by using five state-of-the-art LLMs on our dataset.The results show that most models generate syntactically valid JavaScript object notation(JSON)at high rates(about 80%–99%)and initiate clarifying questions in nearly all incomplete cases(about 90%–100%).However,end-to-end accuracy remains low(about 6%–14%),and missing information is substantial(about 12%–66%).Models also tend to over-interrogate—roughly 70%–80%of follow-ups are unnecessary—raising interaction costs and potential information-exposure risk.The research also demonstrates the feasibility of integrating our pipeline with the real-world trading systems,paving the way for practical deployment of LLM-based trade automation solutions.展开更多
Large language models(LLMs)exhibit remarkable capabilities in various natural language processing tasks,such as machine translation.However,the large number of LLM parameters incurs significant costs during inference....Large language models(LLMs)exhibit remarkable capabilities in various natural language processing tasks,such as machine translation.However,the large number of LLM parameters incurs significant costs during inference.Previous studies have attempted to train translation-tailored LLMs with moderately sized models by fine-tuning them on the translation data.Nevertheless,when performing translations in zero-shot directions that are absent from the fine-tuning data,the problem of ignoring instructions and thus producing translations in the wrong language(i.e.,the off-target translation issue)remains unresolved.In this work,we design a twostage fine-tuning algorithm to improve the instruction-following ability of translation-tailored LLMs,particularly for maintaining accurate translation directions.We first fine-tune LLMs on the translation data to elicit basic translation capabilities.At the second stage,we construct instruction-conficting samples by randomly replacing the instructions with the incorrect ones.Then,we introduce an extra unlikelihood loss to reduce the probability assigned to those samples.Experiments on two benchmarks using the LLaMA 2 and LLaMA 3 models,spanning 16 zero-shot directions,demonstrate that,compared to the competitive baseline translation-finetuned LLaMA,our method could effectively reduce the off-target translation ratio(up to-62.4 percentage points),thus improving translation quality(up to+9.7 bilingual evaluation understudy).Analysis shows that our method can preserve the model's performance on other tasks,such as supervised translation and general tasks.Code is released at https://github.com/alphadl/LanguageAware_Tuning.展开更多
基金supported by the National Natural Science Foundation of China(NSFC)under grant numbers of 52277222,52406256,and 52177217the Shanghai Science and Technology Development Fund under grant number 22ZR14445000the Artificial Intelligence for Research Paradigm Reform Enabling Discipline Leapfrog Program Project Funding Grant。
文摘The explosive growth of lithium-ion battery literature has led to severe knowledge overload,challenging researchers'ability to efficiently extract structured information.While large language models(LLMs)offer considerable potential for automating this task,their practical application in scientific domains is nonetheless constrained by high application programming interface(API)costs and computational resources required for fine-tuning.To address these limitations,a cognition-enhanced instruction framework(CEIF)is proposed,wherein a high-performance teacher model(such as DeepSeek-R1)provides dynamic feedback,prompt refinement,and training data optimization to guide the learning process of low-parameter models.Experimental results demonstrate that the low-parameter models(6B-9B)optimized via the CEIF achieve approximately 85%accuracy in battery literature extraction tasks,approaching the performance of GPT-4 while requiring only a single NVIDIA RTX 3090 GPU.Furthermore,the emergence of an"Aha moment"characterized by rapid performance improvement during specialized learning is observed,offering novel theoretical insights for the design and optimization of domainspecific models.
基金supported by the National Key R&D Program of China[2022YFF0902703]the State Administration for Market Regulation Science and Technology Plan Project(2024MK033).
文摘Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interactions to predict future items of interest.However,many current methods rely on unique user and item IDs,limiting their ability to represent users and items effectively,especially in zero-shot learning scenarios where training data is scarce.With the rapid development of Large Language Models(LLMs),researchers are exploring their potential to enhance recommendation systems.However,there is a semantic gap between the linguistic semantics of LLMs and the collaborative semantics of recommendation systems,where items are typically indexed by IDs.Moreover,most research focuses on item representations,neglecting personalized user modeling.To address these issues,we propose a sequential recommendation framework using LLMs,called CIT-Rec,a model that integrates Collaborative semantics for user representation and Image and Text information for item representation to enhance Recommendations.Specifically,by aligning intuitive image information with text containing semantic features,we can more accurately represent items,improving item representation quality.We focus not only on item representations but also on user representations.To more precisely capture users’personalized preferences,we use traditional sequential recommendation models to train on users’historical interaction data,effectively capturing behavioral patterns.Finally,by combining LLMs and traditional sequential recommendation models,we allow the LLM to understand linguistic semantics while capturing collaborative semantics.Extensive evaluations on real-world datasets show that our model outperforms baseline methods,effectively combining user interaction history with item visual and textual modalities to provide personalized recommendations.
基金Key Scientific Research Project of the Hunan Provincial Department of Education(23A312).
文摘Objective To investigate methods for constructing a high-quality instructional dataset for traditional Chinese medicine(TCM)mental disorders and to validate its efficacy.Methods We proposed the Fine-Med-Mental-T&P methodology for constructing high-quality instruction datasets in TCM mental disorders.This approach integrates theoretical knowledge and practical case studies through a dual-track strategy.(i)Theoretical track:textbooks and guidelines on TCM mental disorders were manually segmented.Initial responses were generated using DeepSeek-V3,followed by refinement by the Qwen3-32B model to align the expression with human preferences.A screening algorithm was then applied to select 16000 high-quality instruction pairs.(ii)Practical track:starting from over 600 real clinical case seeds,diagnostic and therapeutic instruction pairs were generated using DeepSeek-V3 and subsequently screened through manual evaluation,resulting in 4000 high-quality practiceoriented instruction pairs.The integration of both tracks yielded the Med-Mental-Instruct-T&P dataset,comprising a total of 20000 instruction pairs.To validate the dataset’s effectiveness,three experimental evaluations(both manual and automated)were conducted:(i)comparative studies to compare the performance of models fine-tuned on different datasets;(ii)benchmarking to compare against mainstream TCM-specific large language models(LLMs);(iii)data ablation study to investigate the relationship between data volume and model performance.Results Experimental results demonstrate the superior performance of T&P-model finetuned on the Med-Mental-Instruct-T&P dataset.In the comparative study,the T&P-model significantly outperformed the baseline models trained solely on self-generated or purely human-curated baseline data.This superiority was evident in both automated metrics(ROUGEL>0.55)and expert manual evaluations(scoring above 7/10 across accuracy).In benchmark comparisons,the T&P-model also excelled against existing mainstream TCM LLMs(e.g.,HuatuoGPT and ZuoyiGPT).It showed particularly strong capabilities in handling diverse clinical presentations,including challenging disorders such as insomnia and coma,showcasing its robustness and versatility.Data ablation studies showed that T&P-model performance had an overall upward trend with minor fluctuations when training data increased from 10%to 50%;beyond 50%,performance improvement slowed significantly,with metrics plateauing and approaching a saturation point.
文摘This paper firstly discusses five models of bilingual education experimented or applied in different countries of the world.Then it puts forward two feasible models of classroom bilingual instruction—the high level and the low level.Finally it tentatively points out the course criteria of classroom bilingual instruction in Chinese universities,which are believed to be of considerable reference value to bilingual instruction in China.
文摘SPSS statistical analysis system is used to examine the teaching effects of three different instruction models which are commonly used in military English teaching. Of the three models, interactive teaching model has shown to have advantage over the other two, the situational teaching model and the task-based Instruction model. It is concluded that it is beneficial to improve teaching effects in military English teaching if the teacher can stimulate students to involve themselves more in class. To do so, it is necessary to strike a balance between the learners＇ language input and output so as to ensure that the input knowledge be consolidated and digested through practice in output.
文摘The paper presents a case study on listening-speaking class instruction models based on exploratory practice. Assuming the dual roles of a teacher and a researcher, the writer of this paper has implemented three models of classroom instructions for the Viewing, Listening ＆ Speaking Class. From the students＇ report, teacher＇s log and classroom observation, it is concluded that the three models meet the needs of different students. The key to a successful Viewing, Listening＆ Speaking Class is to set reasonable goals and analyze students＇ needs.
文摘Online synchronous educational experience via videoconferencing has become ubiquitous domestically and internationally with the spread of Covid-19 Virus.Online teaching theories and educational experiences through computer-mediated communication have been integrating and evolving during the current situation.The objective of this paper is to demonstrate a practical teaching model for Chinese college students’English as a foreign language learning course.The key finding of the research is that the 5-step-teaching Model developed under the context of online synchronous learning and teaching is both functional and efficient on the theoretical basis of constructivist approach.This teaching model is proved to be practical,interesting,useful and functional to both second-years and graduates during the implementation of educational transaction according to the survey conducted at the end of the semester.The ubiquity of online synchronous learning and teaching would lead to a meaningful and worthwhile educational experience and have a far-reaching implication for both educators and student’s personal development,especially students’critical thinking capability.The findings of the paper are pedagogically valuable for e-teachers in conducting online courses.It also holds true that the management of education and society could benefit and adopt this teaching model to various courses in different institutions.This study is hopefully to provide research-and-experience-based findings to enrich course design and instructional approaches and educational experiences.Thus,it is of interest to online teachers,instructional designers and programme developers in the field of education.
文摘Generative artificial intelligence,represented by large language models,holds vast application scenarios and significant development potential in the field of language teaching.This study employs large language models such as ChatGPT4o,ERNIE Bot,and Spark Cognition to explore how they empower teachers in international Chinese language teaching through practical cases.It focuses on various aspects of international Chinese language teaching and language skills training,examining the application effects of large language models in generating tailored teaching content and converting textual content into multimodal teaching materials.Finally,the study proposes that teachers should rationally recognize the opportunities and challenges that large language models bring to the teaching ecosystem,while acknowledging the models’efficiency in empowering teachers’instruction,it is crucial to fully recognize their essential tool nature,uphold teachers’subjectivity,and pay close attention to the boundaries of their development and application.
文摘We have come to know comprehension better by studying the nature and characteristics of reading processes and by weighing the pros and cons of different reading theories and their influence on the teaching of reading. But in China, unfortunately, few achievements have been made in the field of reading theory research. Thus, in order to ameliorate our reading teaching, it is urgent for us to set up our own system of reading theories. The author, after studying carefully the present reading models in China and by absorbing the quintessence of foreign reading theories, ventures to suggest the “discourse word discourse” model. The model gives full thought to the characteristics of Chinese college students and thus suits a reading classroom in China better. However, the comprehension factors involved in the model are far from exhaustive and the model itself leaves much room for modification and addition.\;
文摘Guided by the"Healthy China 2030"strategy,improving national nutrition and health literacy has become a core task in public health system development.The National Nutrition Plan(2017-2030)explicitly calls for"strengthening the training of nutrition talents"and"promoting nutrition science education".As a key vehicle for this mission,the Food Nutrition and Health course in higher education urgently needs to address bottlenecks in traditional teaching,such as low knowledge application and transfer rates,insufficient student engagement,and ineffective guidance on healthy behaviors.The BOPPPS teaching model,with its structured design(Bridge-in,Objective,Pre-assessment,Participatory Learning,Post-assessment,Summary),effectively promotes the internalization of nutritional knowledge and the transformation into healthy behaviors among students by emphasizing practice-oriented teaching activities.In this study,focusing on this course,an in-depth exploration of curriculum teaching design was conducted based on the BOPPPS instructional model,aiming to deeply integrate the strategic objectives of Healthy China into the curriculum,and promote the transformation of nutritional knowledge into healthy decision-making ability.This study provides new insights for food and nutrition education.
文摘From 2010,a drastic increase in international students’enrollment in many higher education institutions in China has led to the growth of English-medium instruction(EMI)practices and relevant research.With many state-level and province-level initiatives to promote the design and development of model EMI courses,international students’perspectives of these courses are not adequately reflected in the international literature which is primarily done in European higher education contexts.This study takes a qualitative approach by analyzing international students’EMI courses reflection reports,focus group interviews and observation notes of video-recorded EMI courses to probe into their perspectives on EMI practices in a business institute in a university of Shanghai.The findings show that international students show concerns about the systematic design of the curriculum,pedagogical problems and contextual limitations in EMI practices.All EMI stakeholders like curriculum developers,teaching administrators,teachers and students are responsible for these problems.Therefore,meaningful and consistent communication should be greatly promoted among all EMI practices stakeholders.The article concludes with recommendations for further research on model EMI courses in China.
文摘As we have entered 21st century,new technology has had a great impact on educational instructions. In this paper,I will demonstrate how new technology influences the educational instructions. Then,I try to construct a new instruction model——"classroom+internet"model. I will explain the characteristics of new technology and its impact on education. The final part is summary.
文摘The study describes the improvements of the participants in ESP course incorporating Sheltered Instruction Observation Protocol(SIOP) instruction in China. It aims to investigate whether or not it is effective to incorporate the sheltered instruction into ESP course and whether or not the participants will be motivated to improve their English skills for professional purposes. Quantitative and qualitative data sources were employed, including English language tests, questionnaires, and interviews.Findings reveal that the sheltered instruction was helpful to ESP course and produced significant achievements in participants' English vocabulary and skills. The most interesting result is that the participants' confidence and interest in learning professional English for future jobs has been improved much.
基金the Technology Project of Henan Province(No.202102310210)the Key Project of Discipline Construction of Zhengzhou University(No.XKZDQY201905)。
文摘Instruction cues are widely employed for research on neural mechanisms during movement preparation.However,their influence on brain connectivity during movement has not received much attention.Herein,15 healthy subjects completed two experimental tasks including either instructed or voluntary movements;meanwhile electroencephalogram(EEG)data were synchronously recorded.Based on source analysis and related literature,six movement-related brain regions were selected,including the left/right supplementary motor area(SMA),left/right inferior frontal gyrus(iFg),and left/right postcentral gyrus(pCg).After assuming 10 a priori models of regional brain connectivity,we evaluated the optimal connectivity model between brain regions for the two scenarios using the dynamic causality model(DCM).During voluntary movement,the movement originated in the SMA,passed through the iFg of the prefrontal lobe,and then returned to the main sensory cortex of the pCg.In the instructed movement,the movement originated in the iFg,and then was transmitted to the pCg and the SMA,as well as from the pCg to the SMA.In contrast to the preparation process of voluntary movement,there were long-range information interactions between the iFg and pCg.Further,almost the same brain regions were active during movement preparation under both voluntary and instructed movement tasks,which evidences certain similarities in dynamic brain connectivity,that is,the brain has direct connections between the bilateral SMA,bilateral pCg,and bilateral SMA,indicating that the both brain hemispheres work together during the movement preparation phase.The results suggest that the network during the preparation process of instructed movements is more complex than voluntary movements.
基金supported by National Natural Science Foundation of China(No.62261023)National Natural Science Foundation of China(No.U1836118)Science and Technology Innovation 2030“New Generation of Artificial Intelligence”(2020AAA0108501).
文摘Instruction fine-tuning is a key method for adapting large language models(LLMs)to domain-specific tasks,and instruction quality significantly impacts model performance after fine-tuning.Hence,evaluating the quality of instruction and selecting high-quality instructions are essential steps in the process of LLM instruction fine-tuning.Although existing studies provide important theoretical foundations and techniques for this,there is still room for improvement in terms of generality,the relationship between methods and experimental verification.Current methods for evaluating instruction quality can be classified into four main categories:human evaluation,statistics-based evaluation,model-based evaluation,and LLMs-based evaluation.Among these methods,human evaluation relies on the subjective judgment and domain expertise of the evaluators,which offers interpretability and is suitable for scenarios involving small-scale data and sufficient budgets.Statistics-based evaluation estimates the quality of instructions using indicators such as stopwords and lexical diversity,providing high efficiency and a suitable evaluation for large-scale data.Model-based evaluation employs specific models to quantify indicators such as perplexity(PPL)and instruction following difficulty(IFD),which is flexible and suitable for specific tasks.The LLMs-based evaluation rates the quality of instructions through prompt-based interaction with LLMs,focusing on aspects such as accuracy and coherence,which is highly automated and customizable,simplifying the evaluation process.Finally,considering the limitations of current quality evaluation methods,some future research directions are proposed for improvement.These include refining instruction categories,extending evaluation indicators,enhancing human-AI interaction evaluation method,applying agents in instruction quality evaluation,and developing a comprehensive evaluation framework.
文摘The development of large language models(LLMs)has created transformative opportunities for the financial industry,especially in the area of financial trading.However,how to integrate LLMs with trading systems has become a challenge.To address this problem,we propose an intelligent trade order recognition pipeline that enables the conversion of trade orders into a standard format for trade execution.The system improves the ability of human traders to interact with trading platforms while addressing the problem of misinformation acquisition in trade execution.In addition,we create a trade order dataset of 500 pieces of data to simulate the real-world trading scenarios.Moreover,we design several metrics to provide a comprehensive assessment of dataset reliability and the generative power of big models in finance by using five state-of-the-art LLMs on our dataset.The results show that most models generate syntactically valid JavaScript object notation(JSON)at high rates(about 80%–99%)and initiate clarifying questions in nearly all incomplete cases(about 90%–100%).However,end-to-end accuracy remains low(about 6%–14%),and missing information is substantial(about 12%–66%).Models also tend to over-interrogate—roughly 70%–80%of follow-ups are unnecessary—raising interaction costs and potential information-exposure risk.The research also demonstrates the feasibility of integrating our pipeline with the real-world trading systems,paving the way for practical deployment of LLM-based trade automation solutions.
基金Project supported by the National Natural Science Foundation of China(No.62372468)the Shandong Natural Science Foundation(No.ZR2023MF008)+1 种基金the Major Basic Research Projects in Shandong Province(No.ZR2023ZD32)the Qingdao Natural Science Foundation(No.23-2-1-161-zyyd-jch)。
文摘Large language models(LLMs)exhibit remarkable capabilities in various natural language processing tasks,such as machine translation.However,the large number of LLM parameters incurs significant costs during inference.Previous studies have attempted to train translation-tailored LLMs with moderately sized models by fine-tuning them on the translation data.Nevertheless,when performing translations in zero-shot directions that are absent from the fine-tuning data,the problem of ignoring instructions and thus producing translations in the wrong language(i.e.,the off-target translation issue)remains unresolved.In this work,we design a twostage fine-tuning algorithm to improve the instruction-following ability of translation-tailored LLMs,particularly for maintaining accurate translation directions.We first fine-tune LLMs on the translation data to elicit basic translation capabilities.At the second stage,we construct instruction-conficting samples by randomly replacing the instructions with the incorrect ones.Then,we introduce an extra unlikelihood loss to reduce the probability assigned to those samples.Experiments on two benchmarks using the LLaMA 2 and LLaMA 3 models,spanning 16 zero-shot directions,demonstrate that,compared to the competitive baseline translation-finetuned LLaMA,our method could effectively reduce the off-target translation ratio(up to-62.4 percentage points),thus improving translation quality(up to+9.7 bilingual evaluation understudy).Analysis shows that our method can preserve the model's performance on other tasks,such as supervised translation and general tasks.Code is released at https://github.com/alphadl/LanguageAware_Tuning.
