Ahnj-Stability commited on
Commit
37dd96b
·
1 Parent(s): 7e3b8dc

fix: 줄바꿈

Browse files
Files changed (2) hide show
  1. README.md +4 -0
  2. README_en.md +4 -0
README.md CHANGED
@@ -26,18 +26,22 @@ AI 에이전트의 능력이 고도화되면서, 그 성능을 실제 환경과
26
 
27
  # Ko-AgentBench 핵심 특징 ✨
28
  **1. 단계별 태스크 설계**
 
29
  단순 도구 호출부터 장기적 맥락 능력, 강건성 처리 능력까지 에이전트의 능력을 7단계로 입체적으로 분석하였습니다.
30
 
31
  **2. 18가지 한국형 API 사용 및 실생활 환경에 특화된 고품질 시나리오 구성**
 
32
  네이버, 지도, 카카오, 웹사이트 등 한국 실사용 환경 기반의 API를 기반으로 국내 사용자의 일상과 밀접한 '약속 예약', '블로그 후기 검색'과 같은 현실적인 문제 해결 시나리오를 구현했습니다.
33
 
34
  **3. 캐시 기반 반복 평가 및 강건성 테스트**
 
35
  '정보 속성 불일치성 변경' 등 기존 벤치마크의 고질적 문제를 해결합니다.
36
  실패 API 응답을 개선함에 따라 벤치마크의 일관성과 신뢰도를 보장합니다.
37
 
38
  의도된 오류 상황에서의 오류 인식/대응 능력(전략)까지 평가함으로 현실 환경에서도 안정적으로 작동하는 모델을 선별합니다.
39
 
40
  **4. 단계별 고유 정밀 지표**
 
41
  도구 선택, 파라미터 구성, 데이터 흐름 등 문제 해결의 불필요/소요를 단계별로 평가합니다. 이를 통해 모델의 강/약점 정량적으로 식별합니다.
42
 
43
  ## **데이터 로드**
 
26
 
27
  # Ko-AgentBench 핵심 특징 ✨
28
  **1. 단계별 태스크 설계**
29
+
30
  단순 도구 호출부터 장기적 맥락 능력, 강건성 처리 능력까지 에이전트의 능력을 7단계로 입체적으로 분석하였습니다.
31
 
32
  **2. 18가지 한국형 API 사용 및 실생활 환경에 특화된 고품질 시나리오 구성**
33
+
34
  네이버, 지도, 카카오, 웹사이트 등 한국 실사용 환경 기반의 API를 기반으로 국내 사용자의 일상과 밀접한 '약속 예약', '블로그 후기 검색'과 같은 현실적인 문제 해결 시나리오를 구현했습니다.
35
 
36
  **3. 캐시 기반 반복 평가 및 강건성 테스트**
37
+
38
  '정보 속성 불일치성 변경' 등 기존 벤치마크의 고질적 문제를 해결합니다.
39
  실패 API 응답을 개선함에 따라 벤치마크의 일관성과 신뢰도를 보장합니다.
40
 
41
  의도된 오류 상황에서의 오류 인식/대응 능력(전략)까지 평가함으로 현실 환경에서도 안정적으로 작동하는 모델을 선별합니다.
42
 
43
  **4. 단계별 고유 정밀 지표**
44
+
45
  도구 선택, 파라미터 구성, 데이터 흐름 등 문제 해결의 불필요/소요를 단계별로 평가합니다. 이를 통해 모델의 강/약점 정량적으로 식별합니다.
46
 
47
  ## **데이터 로드**
README_en.md CHANGED
@@ -27,18 +27,22 @@ To address this issue, we have developed a high-quality agent benchmark speciali
27
 
28
  # Ko-AgentBench Key Features ✨
29
  **1. Step-by-step Task Design**
 
30
  We have comprehensively analyzed agent capabilities across 7 levels, from simple tool calls to long-term contextual abilities and robustness handling capabilities.
31
 
32
  **2. 18 Korean-specific APIs and High-quality Scenarios Tailored to Real-life Environments**
 
33
  Based on APIs from Korean real-world usage environments such as Naver, Maps, Kakao, and websites, we have implemented realistic problem-solving scenarios closely related to domestic users' daily lives, such as 'appointment booking' and 'blog review search'.
34
 
35
  **3. Cache-based Iterative Evaluation and Robustness Testing**
 
36
  We solve chronic problems of existing benchmarks, such as 'information attribute inconsistency changes'.
37
  By improving failed API responses, we ensure benchmark consistency and reliability.
38
 
39
  By evaluating error recognition/response capabilities (strategies) in intentional error situations, we select models that operate stably even in real-world environments.
40
 
41
  **4. Step-specific Precision Metrics**
 
42
  We evaluate the necessity/requirements of problem-solving step by step, including tool selection, parameter configuration, and data flow. Through this, we quantitatively identify the strengths and weaknesses of models.
43
 
44
  ## **Data Loading**
 
27
 
28
  # Ko-AgentBench Key Features ✨
29
  **1. Step-by-step Task Design**
30
+
31
  We have comprehensively analyzed agent capabilities across 7 levels, from simple tool calls to long-term contextual abilities and robustness handling capabilities.
32
 
33
  **2. 18 Korean-specific APIs and High-quality Scenarios Tailored to Real-life Environments**
34
+
35
  Based on APIs from Korean real-world usage environments such as Naver, Maps, Kakao, and websites, we have implemented realistic problem-solving scenarios closely related to domestic users' daily lives, such as 'appointment booking' and 'blog review search'.
36
 
37
  **3. Cache-based Iterative Evaluation and Robustness Testing**
38
+
39
  We solve chronic problems of existing benchmarks, such as 'information attribute inconsistency changes'.
40
  By improving failed API responses, we ensure benchmark consistency and reliability.
41
 
42
  By evaluating error recognition/response capabilities (strategies) in intentional error situations, we select models that operate stably even in real-world environments.
43
 
44
  **4. Step-specific Precision Metrics**
45
+
46
  We evaluate the necessity/requirements of problem-solving step by step, including tool selection, parameter configuration, and data flow. Through this, we quantitatively identify the strengths and weaknesses of models.
47
 
48
  ## **Data Loading**