인적 평가 작업의 결과 이해 - Amazon SageMaker

기계 번역으로 제공되는 번역입니다. 제공된 번역과 원본 영어의 내용이 상충하는 경우에는 영어 버전이 우선합니다.

인적 평가 작업의 결과 이해

인적 작업자를 사용하는 모델 평가 작업을 생성할 때 하나 이상의 지표 유형 을 선택했습니다. 작업팀 구성원이 작업자 포털에서 응답을 평가하면 응답이 humanAnswers json 객체에 저장됩니다. 이러한 응답이 저장되는 방법은 작업이 생성될 때 선택한 지표 유형에 따라 달라집니다.

다음 섹션에서는 이러한 차이점을 설명하고 예제를 제공합니다.

JSON 출력 참조

모델 평가 작업이 완료되면 결과가 Amazon S3에 JSON 파일로 저장됩니다. JSON 객체에는 세 개의 상위 수준 노드 humanEvaluationResult, inputRecord및 가 포함되어 있습니다modelResponses. humanEvaluationResult 키는 모델 평가 작업에 할당된 작업팀의 응답을 포함하는 상위 수준 노드입니다. inputRecord 키는 모델 평가 작업이 생성될 때 모델(들)에 제공된 프롬프트가 포함된 상위 수준 노드입니다. modelResponses 키는 모델(들)의 프롬프트에 대한 응답을 포함하는 상위 수준 노드입니다.

다음 표에는 모델 평가 작업의 JSON 출력에서 발견된 키 값 페어가 요약되어 있습니다.

다음 섹션에서는 각 키 값 페어에 대한 보다 세분화된 세부 정보를 제공합니다.

파라미터 설명

flowDefinitionArn

arn:aws:sagemaker:us-west-2:111122223333:flow-definition/flow-definition-name

인적 루프ARN를 생성한 인적 검토 워크플로(흐름 정의)의 입니다.

humanAnswers

선택한 평가 지표와 관련된 JSON 객체 목록입니다. 자세한 내용은 섹션을 참조하세요에서 찾은 키 값 페어 humanAnswers.

작업자 응답을 포함하는 JSON 객체 목록입니다.

humanLoopName

system-generated-hash 시스템에서 40자의 16진수 문자열을 생성했습니다.
inputRecord
"inputRecord": { "prompt": { "text": "Who invented the airplane?" }, "category": "Airplanes", "referenceResponse": { "text": "Orville and Wilbur Wright" }, "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." }] }
입력 데이터 세트의 항목 프롬프트가 포함된 JSON 객체입니다.
modelResponses
"modelResponses": [{ "modelIdentifier": "arn:aws:bedrock:us-west-2::foundation-model/model-id", "text": "the-models-response-to-the-prompt" }]
모델의 개별 응답입니다.
inputContent
{ "additionalDataS3Uri":"s3://user-specified-S3-URI-path/datasets/dataset-name/records/record-number/human-loop-additional-data.json", "evaluationMetrics":[ { "description":"brief-name", "metricName":"metric-name", "metricType":"IndividualLikertScale" } ], "instructions":"example instructions" }

Amazon S3 버킷에서 인적 루프를 시작하는 데 필요한 인적 루프 입력 콘텐츠입니다.

modelResponseIdMap
{ "0": "sm-margaret-meta-textgeneration-llama-2-7b-1711485008-0612", "1": "jumpstart-dft-hf-llm-mistral-7b-ins-20240327-043352" }

에서 각 모델을 나타내는 방법을 설명합니다answerContent.

에서 찾은 키 값 페어 humanEvaluationResult

모델 평가 작업의 출력humanEvaluationResult에서 에 있는 다음 키 값 페어입니다.

와 연결된 키 값 페어는 섹션을 humanAnswers참조하세요에서 찾은 키 값 페어 humanAnswers.

flowDefinitionArn

  • 모델 평가 작업을 완료하는 데 사용되는 흐름 정의ARN의 입니다.

  • 예:arn:aws:sagemaker:us-west-2:111122223333:flow-definition/flow-definition-name

humanLoopName

  • 시스템에서 40자의 16진수 문자열을 생성했습니다.

inputContent

  • 이 키 값은 지표 유형 과 워커 포털에서 워커에게 제공한 지침을 설명합니다.

    • additionalDataS3Uri: 작업자에 대한 지침이 저장되는 Amazon S3의 위치입니다.

    • instructions: 작업자 포털에서 작업자에게 제공한 지침입니다.

    • evaluationMetrics: 지표의 이름과 해당 설명입니다. 키 값은 모델의 응답을 평가하기 위해 작업자에게 제공되는 도구metricType입니다.

modelResponseIdMap

  • 이 키 값 페어는 선택한 모델의 전체 이름과 작업자 선택 사항이 humanAnswers 키 값 페어의 모델에 매핑되는 방법을 식별합니다.

에서 발견된 키 값 페어 inputRecord

다음 항목은 inputRecord 키 값 페어를 설명합니다.

prompt

  • 모델로 전송된 프롬프트의 텍스트입니다.

category

  • 프롬프트를 분류하는 선택적 범주입니다. 모델 평가 중에 작업자 포털의 작업자가 볼 수 있습니다.

  • 예:"American cities"

referenceResponse

  • 평가 중에 작업자가 참조할 기본 진실을 지정하는 데 JSON 사용되는 입력의 선택적 필드입니다.

responses

  • 다른 모델의 응답을 JSON 포함하는 입력의 선택적 필드입니다.

예제 JSON 입력 레코드입니다.

{ "prompt": { "text": "Who invented the airplane?" }, "category": "Airplanes", "referenceResponse": { "text": "Orville and Wilbur Wright" }, "responses": // All inference must come from a single model [{ "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." }] }

에서 찾은 키 값 페어 modelResponses

모델의 응답과 응답을 제공한 모델을 포함하는 키 값 페어의 배열입니다.

text

  • 프롬프트에 대한 모델의 응답입니다.

modelIdentifier

  • 모델의 이름입니다.

에서 찾은 키 값 페어 humanAnswers

모델의 응답과 작업자가 에서 모델을 평가한 방법을 포함하는 키 값 페어 배열

acceptanceTime

  • 작업자가 작업자 포털에서 작업을 수락한 경우.

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 생성될 때 선택한 평가 지표에 따라 달라집니다. 각 작업자 응답 또는 응답은 새 JSON 객체에 기록됩니다.

answerContent

  • evaluationResults 에는 작업자의 응답이 포함되어 있습니다.

    • 선택 버튼을 선택하면 각 작업자의 결과가 로 표시됩니다"evaluationResults": "comparisonChoice".

      metricName: 지표의 이름

      result: JSON 객체는 0 또는 를 사용하여 작업자가 선택한 모델을 나타냅니다1. 모델이 보도록 매핑되는 값을 확인하려면 . modelResponseIdMap

    • Likert 스케일, 비교를 선택하면 각 작업자의 결과가 로 표시됩니다"evaluationResults": "comparisonLikertScale".

      metricName: 지표의 이름입니다.

      leftModelResponseId: 작업자 포털의 왼쪽에 modelResponseIdMap 표시된 항목을 나타냅니다.

      rightModelResponseId: 작업자 포털의 왼쪽에 modelResponseIdMap 표시된 항목을 나타냅니다.

      result: JSON 객체는 0 또는 를 사용하여 작업자가 선택한 모델을 나타냅니다1. 모델이 매핑되어 보려는 값을 확인하려면 modelResponseIdMap

    • 서수 순위를 선택하면 각 작업자의 결과가 로 표시됩니다"evaluationResults": "comparisonRank".

      metricName: 지표의 이름

      result: JSON 객체 배열입니다. 각 모델(modelResponseIdMap)에 대해 작업자는 를 제공합니다rank.

      "result": [{ "modelResponseId": "0", "rank": 1 }, { "modelResponseId": "1", "rank": 1 }]
    • Likert 스케일을 선택하면 단일 모델 응답에 대한 평가가 선택되고 작업자가 에 저장되는 결과가 표시됩니다"evaluationResults": "individualLikertScale". 작업이 생성될 때 metricName 지정된 에 대한 점수를 포함하는 JSON 배열입니다.

      metricName: 지표의 이름입니다.

      modelResponseId: 점수가 매겨진 모델입니다. 모델이 보도록 매핑되는 값을 확인하려면 . modelResponseIdMap

      result: 작업자가 선택한 likert 스케일 값을 나타내는 키 값 페어입니다.

    • 썸 업/다운을 선택하면 작업자의 결과가 JSON 배열 로 저장됩니다"evaluationResults": "thumbsUpDown".

      metricName: 지표의 이름입니다.

      result: true 또는 와 관련된 false metricName. 작업자가 엄지손가락을 올리면 "result" : true.

모델 평가 작업 출력의 예제 출력

다음 JSON 객체는 Amazon S3에 저장된 모델 평가 작업 출력의 예입니다. 각 키 값 페어에 대한 자세한 내용은 섹션을 참조하세요JSON 출력 참조.

명확성을 위해 이 작업에는 두 작업자의 응답만 포함됩니다. 일부 키 값 페어도 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" } ] }