기계 번역으로 제공되는 번역입니다. 제공된 번역과 원본 영어의 내용이 상충하는 경우에는 영어 버전이 우선합니다.
인간 평가 작업의 결과 이해
인간 작업자를 사용하는 모델 평가 작업을 만들 때 하나 이상의 지표 유형을 선택했습니다. 작업 팀 구성원이 작업자 포털에서 응답을 평가하면 응답이
json 객체에 저장됩니다. 이러한 응답이 저장되는 방법은 작업이 만들어질 때 선택한 지표 유형에 따라 달라집니다.humanAnswers
다음 섹션에서는 이러한 차이점을 설명하고 예시를 제공합니다.
JSON 출력 참조
모델 평가 작업이 완료되면 결과가 Amazon S3에 JSON 파일로 저장됩니다. JSON 객체에는 세 개의 상위 수준 노드인 humanEvaluationResult
, inputRecord
, modelResponses
가 포함되어 있습니다. humanEvaluationResult
키는 모델 평가 작업에 할당된 작업 팀의 응답을 포함하는 상위 수준 노드입니다. inputRecord
키는 모델 평가 작업이 만들어질 때 모델에 제공된 프롬프트가 포함된 상위 수준 노드입니다. modelResponses
키는 모델의 프롬프트에 대한 응답을 포함하는 상위 수준 노드입니다.
다음 표에는 모델 평가 작업의 JSON 출력에 있는 키 값 쌍이 요약되어 있습니다.
이어지는 섹션에서는 각 키 값 쌍에 대한 보다 세분화된 세부 정보를 제공합니다.
파라미터 | 예제 | 설명 |
---|---|---|
|
arn:aws:sagemaker:us-west-2:
|
인적 루프를 만드는 데 사용된 인간 검토 워크플로(흐름 정의)의 ARN입니다. |
humanAnswers |
선택한 평가 지표와 관련된 JSON 객체 목록입니다. 자세한 내용은 humanAnswers에서 찾은 키 값 쌍 섹션을 참조하세요. |
작업자 응답이 포함된 JSON 객체의 목록입니다. |
|
system-generated-hash |
시스템에서 40자의 16진수 문자열을 생성했습니다. |
inputRecord |
|
입력 데이터세트의 입력 프롬프트가 포함된 JSON 객체입니다. |
modelResponses |
|
모델의 개별 응답입니다. |
inputContent |
|
Amazon S3 버킷에서 인적 루프를 시작하는 데 필요한 인적 루프 입력 콘텐츠입니다. |
modelResponseIdMap |
|
|
humanEvaluationResult
에서 찾은 키 값 쌍
다음 키 값 쌍은 모델 평가 작업의 출력에서 humanEvaluationResult
에서 찾을 수 있습니다.
humanAnswers
와 연결된 키 값 쌍은 humanAnswers에서 찾은 키 값 쌍 섹션을 참조하세요.
flowDefinitionArn
-
모델 평가 작업을 완료하는 데 사용되는 흐름 정의의 ARN입니다.
-
예:
arn:aws:sagemaker:us-west-2:
111122223333
:flow-definition/flow-definition-name
humanLoopName
-
시스템에서 40자의 16진수 문자열을 생성했습니다.
inputContent
-
이 키 값은 지표 유형과 작업자 포털에서 작업자에게 제공한 지침을 설명합니다.
-
additionalDataS3Uri
: 작업자에 대한 지침이 저장되는 Amazon S3의 위치입니다. -
instructions
: 작업자 포털에서 작업자에게 제공한 지침입니다. -
evaluationMetrics
: 지표의 이름과 해당 설명입니다.metricType
키 값은 모델의 응답을 평가하기 위해 작업자에게 제공되는 도구입니다.
-
modelResponseIdMap
-
이 키 값 쌍은 선택한 모델의 전체 이름과 작업자의 선택이
humanAnswers
키 값 쌍의 모델에 매핑되는 방법을 식별합니다.
inputRecord
에서 찾은 키 값 쌍
다음 항목은 inputRecord
키 값 쌍을 설명합니다.
prompt
-
모델로 전송된 프롬프트의 텍스트입니다.
category
-
프롬프트를 분류하는 선택적 범주입니다. 모델 평가 중에 작업자 포털에서 작업자가 볼 수 있습니다.
-
예:
"American cities"
referenceResponse
-
평가 중에 작업자가 참조할 실제 응답을 지정하는 데 사용되는 입력 JSON의 선택적 필드입니다.
responses
-
다른 모델의 응답을 포함하는 입력 JSON의 선택적 필드입니다.
예시 JSON 입력 레코드입니다.
{ "prompt": { "text": "Who invented the airplane?" }, "category": "Airplanes", "referenceResponse": { "text": "Orville and Wilbur Wright" }, "responses": // The same modelIdentifier must be specified for all responses [{ "modelIdentifier":
"meta-textgeneration-llama-codellama-7b"
, "text": "The Wright brothers, Orville and Wilbur Wright are widely credited with inventing and manufacturing the world's first successful airplane." }] }
modelResponses
에서 찾은 키 값 쌍
모델의 응답과 응답을 제공한 모델을 포함하는 키 값 쌍의 배열입니다.
text
-
프롬프트에 대한 모델의 응답입니다.
modelIdentifier
-
모델의 이름입니다.
humanAnswers
에서 찾은 키 값 쌍
모델의 응답과 작업자가 모델을 평가한 방법을 포함하는 키 값 쌍의 배열입니다.
acceptanceTime
-
작업자가 작업자 포털에서 작업을 수락한 시간입니다.
submissionTime
-
작업자가 응답을 제출한 시간입니다.
timeSpentInSeconds
-
작업자가 작업을 완료하는 데 소요한 시간입니다.
workerId
-
작업을 완료한 작업자의 ID입니다.
workerMetadata
-
이 모델 평가 작업에 할당된 작업 팀에 대한 메타데이터입니다.
answerContent
JSON 배열의 형식
응답 구조는 모델 평가 작업이 만들어질 때 선택한 평가 지표에 따라 달라집니다. 각 작업자 응답 또는 답변은 새 JSON 객체에 기록됩니다.
answerContent
-
evaluationResults
에는 작업자의 응답이 포함되어 있습니다.-
선택 버튼을 선택하면 각 작업자의 결과가
"evaluationResults": "comparisonChoice"
로 표시됩니다.metricName
: 지표의 이름입니다.result
: JSON 객체는0
또는1
을 사용하여 작업자가 선택한 모델을 나타냅니다. 모델이 매핑되는 값을 확인하려면modelResponseIdMap
을 참조하세요. -
리커트 척도, 비교를 선택하면 각 작업자의 결과가
"evaluationResults": "comparisonLikertScale"
로 표시됩니다.metricName
: 지표의 이름입니다.leftModelResponseId
: 작업자 포털의 왼쪽에 표시된modelResponseIdMap
을 나타냅니다.rightModelResponseId
: 작업자 포털의 왼쪽에 표시된modelResponseIdMap
을 나타냅니다.result
: JSON 객체는0
또는1
을 사용하여 작업자가 선택한 모델을 나타냅니다. 모델이 매핑되는 값을 확인하려면modelResponseIdMap
을 참조하세요. -
서수 순위를 선택하면 각 작업자의 결과가
"evaluationResults": "comparisonRank"
로 표시됩니다.metricName
: 지표의 이름입니다.result
: JSON 객체 배열입니다. 각 모델(modelResponseIdMap
)에 대해 작업자는rank
를 제공합니다."result": [{ "modelResponseId": "0", "rank": 1 }, { "modelResponseId": "1", "rank": 1 }]
-
리커트 척도, 단일 모델의 응답 평가를 선택하면 작업자의 결과가
"evaluationResults": "individualLikertScale"
에 저장됩니다. 이는 작업이 만들어질 때 지정된metricName
에 대한 점수를 포함하는 JSON 배열입니다.metricName
: 지표의 이름입니다.modelResponseId
: 점수가 매겨진 모델입니다. 모델이 매핑되는 값을 확인하려면modelResponseIdMap
을 참조하세요.result
: 작업자가 선택한 리커트 척도 값을 나타내는 키 값 쌍입니다. -
추천/반대를 선택하면 작업자의 결과가 JSON 배열
"evaluationResults": "thumbsUpDown"
으로 저장됩니다.metricName
: 지표의 이름입니다.result
:metricName
과 관련된true
또는false
입니다. 작업자가 추천을 선택하면"result" : true
입니다.
-
모델 평가 작업 출력의 예시 출력
다음 JSON 객체는 Amazon S3에 저장된 모델 평가 작업 출력의 예입니다. 각 키 값 쌍에 대한 자세한 내용은 JSON 출력 참조 섹션을 참조하세요.
명확성을 위해 이 작업에는 두 작업자의 응답만 포함되어 있습니다. 일부 키 값 쌍은 가독성을 위해 잘렸을 수 있습니다.
{ "humanEvaluationResult": { "flowDefinitionArn": "arn:aws:sagemaker:
us-west-2
:111122223333
:flow-definition/flow-definition-name
", "humanAnswers": [ { "acceptanceTime": "2024-06-07T22:31:57.066Z", "answerContent": { "evaluationResults": { "comparisonChoice": [ { "metricName": "Fluency", "result": { "modelResponseId": "0" } } ], "comparisonLikertScale": [ { "leftModelResponseId": "0", "metricName": "Coherence", "result": 1, "rightModelResponseId": "1" } ], "comparisonRank": [ { "metricName": "Toxicity", "result": [ { "modelResponseId": "0", "rank": 1 }, { "modelResponseId": "1", "rank": 1 } ] } ], "individualLikertScale": [ { "metricName": "Correctness", "modelResponseId": "0", "result": 2 }, { "metricName": "Correctness", "modelResponseId": "1", "result": 3 }, { "metricName": "Completeness", "modelResponseId": "0", "result": 1 }, { "metricName": "Completeness", "modelResponseId": "1", "result": 4 } ], "thumbsUpDown": [ { "metricName": "Accuracy", "modelResponseId": "0", "result": true }, { "metricName": "Accuracy", "modelResponseId": "1", "result": true } ] } }, "submissionTime": "2024-06-07T22:32:19.640Z", "timeSpentInSeconds": 22.574, "workerId": "ead1ba56c1278175", "workerMetadata": { "identityData": { "identityProviderType": "Cognito", "issuer": "https://cognito-idp.us-west-2.amazonaws.com/us-west-2_WxGLvNMy4", "sub": "cd2848f5-6105-4f72-b44e-68f9cb79ba07" } } }, { "acceptanceTime": "2024-06-07T22:32:19.721Z", "answerContent": { "evaluationResults": { "comparisonChoice": [ { "metricName": "Fluency", "result": { "modelResponseId": "1" } } ], "comparisonLikertScale": [ { "leftModelResponseId": "0", "metricName": "Coherence", "result": 1, "rightModelResponseId": "1" } ], "comparisonRank": [ { "metricName": "Toxicity", "result": [ { "modelResponseId": "0", "rank": 2 }, { "modelResponseId": "1", "rank": 1 } ] } ], "individualLikertScale": [ { "metricName": "Correctness", "modelResponseId": "0", "result": 3 }, { "metricName": "Correctness", "modelResponseId": "1", "result": 4 }, { "metricName": "Completeness", "modelResponseId": "0", "result": 1 }, { "metricName": "Completeness", "modelResponseId": "1", "result": 5 } ], "thumbsUpDown": [ { "metricName": "Accuracy", "modelResponseId": "0", "result": true }, { "metricName": "Accuracy", "modelResponseId": "1", "result": false } ] } }, "submissionTime": "2024-06-07T22:32:57.918Z", "timeSpentInSeconds": 38.197, "workerId": "bad258db224c3db6", "workerMetadata": { "identityData": { "identityProviderType": "Cognito", "issuer": "https://cognito-idp.us-west-2.amazonaws.com/us-west-2_WxGLvNMy4", "sub": "84d5194a-3eed-4ecc-926d-4b9e1b724094" } } } ], "humanLoopName": "a757 11d3e75a 8d41f35b9873d 253f5b7bce0256e", "inputContent": { "additionalDataS3Uri": "s3://mgrt-test-us-west-2/test-2-workers-2-model/datasets/custom_dataset/0/task-input-additional-data.json", "instructions": "worker instructions provided by the model evaluation job administrator", "evaluationMetrics": [ { "metricName": "Fluency", "metricType": "ComparisonChoice", "description": "Measures the linguistic quality of a generated text." }, { "metricName": "Coherence", "metricType": "ComparisonLikertScale", "description": "Measures the organization and structure of a generated text." }, { "metricName": "Toxicity", "metricType": "ComparisonRank", "description": "Measures the harmfulness of a generated text." }, { "metricName": "Accuracy", "metricType": "ThumbsUpDown", "description": "Indicates the accuracy of a generated text." }, { "metricName": "Correctness", "metricType": "IndividualLikertScale", "description": "Measures a generated answer's satisfaction in the context of the question." }, { "metricName": "Completeness", "metricType": "IndividualLikertScale", "description": "Measures a generated answer's inclusion of all relevant information." } ], "disableRandomization": "true" }, "modelResponseIdMap": { "0": "sm-margaret-meta-textgeneration-llama-2-7b-1711485008-0612", "1": "jumpstart-dft-hf-llm-mistral-7b-ins-20240327-043352" } }, "inputRecord": { "prompt": { "text": "What is high intensity interval training?" }, "category": "Fitness", "referenceResponse": { "text": "High-Intensity Interval Training (HIIT)" } }, "modelResponses": [ { "text": "High Intensity Interval Training (HIIT) is a form of exercise that alternates between periods of high intensity work and low intensity recovery.HIIT is an excellent way to increase your fitness and improve your health, but it can be difficult to get started.In this article, we will", "modelIdentifier": "sm-margaret-meta-textgeneration-llama-2-7b-1711485008-0612" }, { "text": "High intensity interval training is a form of exercise consisting of short bursts of maximum effort followed by periods of rest. The theory behind HIIT is that it can often be more effective at improving cardiovascular and metabolic health than longer, lower intensity workouts.The work intervals can range in length depending on the specific type of exercise, but are typically between 20 and 90 seconds. The recovery periods are generally longer, lasting between 1 and 5 minutes. This pattern is then repeated for multiple sets.\n\nSince the work intervals are high intensity, they require more effort from your body and therefore result in a greater calorie burn. The body also continues to burn calories at an increased rate after the workout due to an effect called excess post exercise oxygen consumption (EPOC), also know as the afterburn effect.\n\nHIIT is a versatile form of training that can be adapted to different fitness levels and can be performed using a variety of exercises including cycling, running, bodyweight movements, and even swimming. It can be done in as little as 20 minutes once or twice a week, making it an efficient option for busy individuals.\n\nWhat are the benefits of high intensity interval training", "modelIdentifier": "jumpstart-dft-hf-llm-mistral-7b-ins-20240327-043352" } ] }