• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.541-544
• /
• 2023

LLM(Largescale Language Model)의 성능 향상을 위한 비용 효율적인 방법으로 ChatGPT, GPT-4와 같은 초거대 모델의 output에 대해 SLM(Small Language Model)을 finetune하는 방법이 주목받고 있다. 그러나, 이러한 접근법은 주로 범용적인 지시사항 모델을 위한 학습 방법으로 사용되며, 제한된 특정 도메인에서는 추가적인 성능 개선의 여지가 있다. 본 연구는 특정 도메인(Writing Assistant)에서의 성능 향상을 위한 새로운 방법인 Self-Guided Approach를 제안한다. Self-Guided Approach는 (1) LLM을 활용해 시드 데이터에 대해 도메인 특화된 metric(유용성, 관련성, 정확성, 세부사항의 수준별) 점수를 매기고, (2) 점수가 매겨진 데이터와 점수가 매겨지지 않은 데이터를 모두 활용하여 supervised 방식으로 SLM을 미세 조정한다. Vicuna에서 제안된 평가 방법인, GPT-4를 활용한 자동평가 프레임워크를 사용하여 Self-Guided Approach로 학습된 SLM의 성능을 평가하였다. 평가 결과 Self-Guided Approach가 Self-instruct, alpaca와 같이, 생성된 instruction 데이터에 튜닝하는 기존의 훈련 방법에 비해 성능이 향상됨을 확인했다. 다양한 스케일의 한국어 오픈 소스 LLM(Polyglot1.3B, PolyGlot3.8B, PolyGlot5.8B)에 대해서 Self-Guided Approach를 활용한 성능 개선을 확인했다. 평가는 GPT-4를 활용한 자동 평가를 진행했으며, Korean Novel Generation 도메인의 경우, 테스트 셋에서 4.547점에서 6.286점의 성능 향상이 발생했으며, Korean scenario Genration 도메인의 경우, 테스트 셋에서 4.038점에서 5.795 점의 성능 향상이 발생했으며, 다른 유사 도메인들에서도 비슷한 점수 향상을 확인했다. Self-Guided Approach의 활용을 통해 특정 도메인(Writing Assistant)에서의 SLM의 성능 개선 가능성을 확인했으며 이는 LLM에 비용부담을 크게 줄이면서도 제한된 도메인에서 성능을 유지하며, LLM을 활용한 응용 서비스에 있어 실질적인 도움을 제공할 수 있을 것으로 기대된다.

• Korean Journal of Plant Resources
• /
• v.8 no.3
• /
• pp.275-280
• /
• 1995

To determine the effect of garlic on acute liver damage caused by $CCl_4$, the values of GOT, GPT, LDH in blood were measured. GOT, GPT, LDH values from mouse treated 2,000mg/kg garlic substance powder were significantly lower than disitlled mouse(positive treatment) after distilled water ingestion only showed big death, but 2,000mg/kg garlic powder treated mouse rarely showed death of liver tissue.

• Korean Journal of Veterinary Service
• /
• v.20 no.2
• /
• pp.177-182
• /
• 1997

To make index on therapy or releasing from dairy holstein cows of reproductive disorder, the levels of serological sex hormones and total ALP, GOT, GPT, and serum protein were measured. The results obtained through the experiments are summarized as follows ; 1. The levels of FSH and estradiol in reproductive disorder group were significantly higher and lower than those in pregnant group and control group, respectively. (p<0.05) 2. The level of prolactin was showed mild difference in three groups and that of progesterone was reasonably higher in pregnant group with significance (p<0.05). 3. The activity of GPT was significantly higher and that of GOT was showed higher In reproductive disorder group, respectively ( p<0.05) 4. The level of total ALP was higher in reproductive disorder group, however no difference was showed between pregnant and control group. 5. The level of serum protein was followed as control, pregnant, and reproductive disorder with insignificance.

• Journal of the East Asian Society of Dietary Life
• /
• v.7 no.2
• /
• pp.181-198
• /
• 1997

The purpose of this study was to investigated relationship between heath status and food preference of male and fermale adults who live in Iksan City, Cheonbuk, Korea. The subjects consisted of 96 male and 93 fermale adults were aged 20 to 70 years old. Each subject was interviewed to get information of food preference. Blood samples were taken concentration of serum in hemoglobin, glucose, total cholesterol, GOT(glutamic oxaloacetic transaminase), GPT(glutamic pyruvic transaminase) were measured. The results obtained were as follows: Among the food preference were like, dislike and ordinary answered to male and fermale adults the normal average of serum concentration with hemoglobin level showed 15.4, 14.2, 15.5, 12.9, 15.8 and 13.2g/dl, glucose level showed 85.8, 86.1, 87.5, 88.1, 87.9 and 86.1mg/dl, total cholesterol level showed 183.1, 185.0, 172.4, 193.5, 181.2 and 184.0mg/dl, GOT level showed 4.8, 23.4, 24.8, 23.9, 24.9 and 21.7ppm, GPT level showed 22.7, 20.2, 26.3, 18.5, 22.5 and 18.4ppm respectively. The abnormal average of serum concentration with hemoglobin level showed 11.1, 10.8, 12.2, 11.3, 12.5 and 11.0g/dl, abnormal glucose level showed 155.7 168.5, 166.2, 134.1, 124.1, 130.1, abnormal total cholesterol level showed 260.3, 273.7, 255.2, 286.5, 255.9 and 251.8mg/dl, abnormal GOT level showed 58.8, 66.8, 51.8, 50.3, 51.2 and 51.0ppm abnormal GPT level showed 54.3, 48.6, 51.3, 50.2, 53.2 and 45.5ppm respectively.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.19 no.2
• /
• pp.371-382
• /
• 2024

This study explored the factors affecting the intention to use ChatGPT for adult men and women. The results are summarized and presented based on the hypothesis as follows. First, perceived ease of use had a positive effect on perceived usefulness. Second, information reliability did not have a significant effect on perceived usefulness. Third, information quickness had a positive effect on perceived usefulness. Fourth, perceived ease of use had a positive effect on user satisfaction. Fifth, perceived usefulness had a positive effect on user satisfaction. Sixth, information reliability did not have a significant effect on user satisfaction. Seventh, information quickness had a positive effect on user satisfaction. Eighth, user satisfaction had a positive effect on the intention to use.

• Journal of Acupuncture Research
• /
• v.25 no.6
• /
• pp.67-76
• /
• 2008

Objectives : The 7-zone-diagnostic system is a diagnostic device to predetermine bodily locations by measuring the energy of body. This study was to investigate the relation between the different patterns of Zone 1, 2, 3 of Factor AA in CP-6000A(VEGA, Germany), 7-zone-diagnostic system and clinical parameters. The purpose of this study was relation Korean traditional medicine and western medicine with the data from 7-zone-diagnostic system and the clinical parameters. Methods : This study was carried out with the data from some clinical parameters. We made three groups according to the Factor AA patterns of CP-6000A. The Factor AA pattern of Group A is that the red bar graph of zone 1, 2, 3 were higher than the normal range and the others were the normal range. The Factor AA pattern of Group B was that the red bar graph of zone 1, 2, 3 was the normal range and the others were the normal range. The Factor AA pattern of Group C was that the red bar graph of zone 1, 2, 3 was lower than the normal range and the others were the normal range. After the data from clinical parameters to correspond with conditions of each group were selected, the data from clinical parameters among each groups analyzed statistically. Results : The values of GOT, GPT, r-GPT, Triglyceride, BUN, Uric acid of group A was higher than group C. Gastroscope of group A and B was higher than group C. Conclusions : It is thought that the red bar graph of zone 1, 2, 3 is higher, the group has the higher energy and the energy has a character of fire(熱). Those patterns have a high risk of hyperlipermia and liver, stomach disease.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.34 no.2
• /
• pp.207-216
• /
• 2024

As big data was built due to the 4th Industrial Revolution, personalized services increased rapidly. As a result, the amount of personal information collected from online services has increased, and concerns about users' personal information leakage and privacy infringement have increased. Online service providers provide privacy policies to address concerns about privacy infringement of users, but privacy policies are often misused due to the long and complex problem that it is difficult for users to directly identify risk items. Therefore, there is a need for a method that can automatically check whether the privacy policy is safe. However, the safety verification technique of the conventional blacklist and machine learning-based privacy policy has a problem that is difficult to expand or has low accessibility. In this paper, to solve the problem, we propose a safety verification technique for the privacy policy using the GPT-3.5 API, which is a generative artificial intelligence. Classification work can be performed evenin a new environment, and it shows the possibility that the general public without expertise can easily inspect the privacy policy. In the experiment, how accurately the blacklist-based privacy policy and the GPT-based privacy policy classify safe and unsafe sentences and the time spent on classification was measured. According to the experimental results, the proposed technique showed 10.34% higher accuracy on average than the conventional blacklist-based sentence safety verification technique.

• Toxicological Research
• /
• v.7 no.2
• /
• pp.165-171
• /
• 1991

Effects of forced physical exercise on the CCl4 hepatotoxicity were examined in adult female rats. Rats were treated with CCl4 (2 mmol/kg, ip) and introduced into a cylindrical cage rotating at 9 rpm for 20 min/hr for 6 hr. Eighteen hr following the termination of exercise serum sorbitol dehydrogenase (SDH), glutamic oxaloacetic transaminase (GOT), glutamic pyruvic transaminase (GPT) activities and glucose-6-phosphatase activity were determined as parameters for hepatotoxicity. Physical exercise inhibited the $CCI_{4-}$induced increases in SDH, GOT, GPT activity, and glucose-6-phosphatase activity were determined as parameters for hepatotoxicity.

• The Journal of Korean Medicine
• /
• v.17 no.2 s.32
• /
• pp.214-226
• /
• 1996

In an attempt to evaluate the effect of high and low concentration of Injinsaryungsan and high and low concentration of Sosihotang on cholestatic liver injery induced by $ANIT({\alpha}-naphthylisothiocyanate)$, biochemical changes in serum transaminase(GOT, GPT), alkaline phosphate, lactate dehydrogenase, total cholesterol, triglyceride, total-bilirubine were studied and the following results were obtained. 1. High concentration of Injinsaryungsan(2.2g／Kg) inhibited significantly the activity increases of GOT, GPT, ALP, LDH, TC, TG, T-Bilirubine induced by $ANIT({\alpha}-naphthylisothiocyanate)$. 2. Low concentration of Injinsaryungsan(1.1g／Kg) inhibited the activity increases of ALP, LDH, TC, TG with statistical significance, while inhibited the activity increase of GOT ,but with no statistical significance. 3. High concentration of Sosihotang(2.4g／Kg) inhibited the activity increases of LDH, TG, TC with statistical significance while inhibited the activity increases of GOT, GPT, ALP, T-bilirubine with no significance. 4. Low concentration of Sosihotang(1.2g／Kg) inhibited the activity increase of TG, while inhibited the activity increase of ALP, TC with no statistical sig-nificance, but didn't inhibite the activity increases of GOT, GPT, LDH, T-Bil. These results suggest that Injinsaryungsan has more significant effect on the liver injury induced by $ANIT({\alpha}-naphthylisothiocyanate)$ compared with Sosihotang and so can be applicable clinically to virus hepatitis and cholestatic liver injury. Further study will be required to evaluate the effect of Sosibotang on cholangitis and cholecystitis.

• Research in Plant Disease
• /
• v.30 no.1
• /
• pp.99-102
• /
• 2024

Integrated pest management is essential for controlling plant diseases that reduce crop yields. Rapid diagnosis is crucial for effective management in the event of an outbreak to identify the cause and minimize damage. Diagnosis methods range from indirect visual observation, which can be subjective and inaccurate, to machine learning and deep learning predictions that may suffer from biased data. Direct molecular-based methods, while accurate, are complex and time-consuming. However, the development of large multimodal models, like GPT-4, combines image recognition with natural language processing for more accurate diagnostic information. This study introduces GPT-4-based system for diagnosing plant diseases utilizing a detailed knowledge base with 1,420 host plants, 2,462 pathogens, and 37,467 pesticide instances from the official plant disease and pesticide registries of Korea. The AI plant doctor offers interactive advice on diagnosis, control methods, and pesticide use for diseases in Korea and is accessible at https://pdoc.scnu.ac.kr/.