Terjemahan disediakan oleh mesin penerjemah. Jika konten terjemahan yang diberikan bertentangan dengan versi bahasa Inggris aslinya, utamakan versi bahasa Inggris.

Menerapkan fungsi hadiah

Ikhtisar

Fungsi hadiah (juga disebut pencetak gol atau grader) adalah komponen inti yang mengevaluasi respons model dan memberikan sinyal umpan balik untuk pelatihan. Ini harus diimplementasikan sebagai fungsi Lambda yang menerima respons model dan mengembalikan skor hadiah.

Format antarmuka

Fungsi reward Anda harus menerima dan mengembalikan data dalam format berikut:

Sampel sampel masukan untuk pelatihan

{  
    "messages": [  
        {  
            "role": "user",  
            "content": "Do you have a dedicated security team?"  
        }  
    ],              
   "reference_answer": {  
       "compliant": "No",  
       "explanation": "As an AI developed by Company, I do not have a traditional security team..."  
    }  
}

Contoh payload untuk hadiah lambda

Kontainer secara otomatis mengubah data Anda sebelum mengirimnya ke fungsi Lambda Anda dengan:

Fungsi Lambda Anda akan menerima data dalam format yang diubah ini:

{    
   "id": "123",  
    "messages": [  
        {  
            "role": "user",  
            "content": "Do you have a dedicated security team?"  
        },  
        {  
            "role": "assistant",  
            "content": "As an AI developed by Amazon, I don not have a dedicated security team..."  
        }  
    ],              
    # Following section will be same as your training dataset sample  
    "reference_answer": {  
        "compliant": "No",  
        "explanation": "As an AI developed by Company, I do not have a traditional security team..."  
    }  
}

Hadiah kontrak Lambda

def lambda_handler(event, context):  
   return lambda_grader(event)  
  
def lambda_grader(samples: list[dict]) -> list[dict]:  
    """  
    Args:  
        samples: List of dictionaries in OpenAI format  
          
        Example input:  
        {     
            "id": "123",  
            "messages": [  
                {  
                    "role": "user",  
                    "content": "Do you have a dedicated security team?"  
                },  
                {  
                    "role": "assistant",  
                    "content": "As an AI developed by Company, I don nott have a dedicated security team..."  
                }  
            ],              
            # This section will be same as your training dataset  
            "reference_answer": {  
                "compliant": "No",  
                "explanation": "As an AI developed by Company, I do not have a traditional security team..."  
            }  
        }  
      
    Returns:  
        List of dictionaries with reward scores:  
        {  
            "id": str,                              # Same id as input sample  
            "aggregate_reward_score": float,        # Overall score for the sample  
            "metrics_list": [                       # OPTIONAL: Component scores  
                {  
                    "name": str,                    # Name of the component score  
                    "value": float,                 # Value of the component score  
                    "type": str                     # "Reward" or "Metric"  
                }  
            ]  
        }  
    """

Bidang input dan output

Bidang input

Bidang output

Kendala teknis

Praktik terbaik

Menggunakan fungsi hadiah khusus

Menerapkan fungsi hadiah khusus ketika Anda memiliki kriteria evaluasi khusus tugas:

Izin IAM

Izin yang diperlukan

Peran SageMaker eksekusi Anda harus memiliki izin untuk menjalankan fungsi Lambda Anda. Tambahkan kebijakan ini ke peran SageMaker eksekusi Anda:

{  
  "Version": "2012-10-17",		 	 	   
  "Statement": [  
    {  
      "Effect": "Allow",  
      "Action": [  
        "lambda:InvokeFunction"  
      ],  
      "Resource": "arn:aws:lambda:region:account-id:function:function-name"  
    }  
  ]  
}

Peran pelaksanaan Lambda

Peran eksekusi fungsi Lambda Anda memerlukan izin eksekusi Lambda dasar:

{  
  "Version": "2012-10-17",		 	 	   
  "Statement": [  
    {  
      "Effect": "Allow",  
      "Action": [  
        "logs:CreateLogGroup",  
        "logs:CreateLogStream",  
        "logs:PutLogEvents"  
      ],  
      "Resource": "arn:aws:logs:*:*:*"  
    }  
  ]  
}

Izin tambahan: Jika fungsi Lambda Anda mengakses layanan AWS lain (misalnya, S3 untuk data referensi, DynamoDB untuk logging), tambahkan izin tersebut ke peran eksekusi Lambda.

Contoh: LLM Sebagai fungsi hadiah Hakim

Contoh ini menunjukkan penggunaan model Amazon Bedrock sebagai juri untuk mengevaluasi respons model dengan membandingkannya dengan jawaban referensi. Template Lambda ini menyediakan kerangka kerja bagi pelanggan untuk mengimplementasikan panggilan ke Amazon Bedrock untuk permintaan inferensi guna memproses evaluasi penilaian. Fungsi Lambda mempertahankan input/output kontrak yang sama dengan fungsi hadiah lainnya.

Implementasi

Fungsi Lambda ini mengimplementasikan proses evaluasi dua tahap: lambda_handler mengekstrak respons model dan jawaban referensi dari sampel yang masuk, kemudian fungsi tersebut lambda_graded memanggil Amazon Bedrock untuk menilai kesamaan semantik di antara mereka. Implementasinya mencakup penanganan kesalahan yang kuat dengan percobaan ulang otomatis untuk kegagalan sementara dan mendukung format jawaban referensi yang fleksibel (baik format string dan kamus terstruktur).

Detail implementasi:

"""  
LLM Judge Lambda POC - Working implementation using Amazon Bedrock  
"""  
  
import json  
import time  
import boto3  
  
bedrock_runtime = boto3.client('bedrock-runtime', region_name='us-east-1')  
JUDGE_MODEL_ID = "anthropic.claude-3-5-sonnet-20240620-v1:0"  
SYSTEM_PROMPT = "You must output ONLY a number between 0.0 and 1.0. No explanations, no text, just the number."  
  
JUDGE_PROMPT_TEMPLATE = """Compare the following two responses and rate how similar they are on a scale of 0.0 to 1.0, where:  
- 1.0 means the responses are semantically equivalent (same meaning, even if worded differently)  
- 0.5 means the responses are partially similar  
- 0.0 means the responses are completely different or contradictory  
  
Response A: {response_a}  
  
Response B: {response_b}  
  
Output ONLY a number between 0.0 and 1.0. No explanations."""  
  
  
def lambda_graded(response_a: str, response_b: str, max_retries: int = 3) -> float:  
    """Call Bedrock to compare responses and return similarity score."""  
    prompt = JUDGE_PROMPT_TEMPLATE.format(response_a=response_a, response_b=response_b)  
      
    for attempt in range(max_retries):  
        try:  
            response = bedrock_runtime.converse(  
                modelId=JUDGE_MODEL_ID,  
                messages=[{"role": "user", "content": [{"text": prompt}]}],  
                system=[{"text": SYSTEM_PROMPT}],  
                inferenceConfig={"temperature": 0.0, "maxTokens": 10}  
            )  
            print(f"Bedrock call successful: {response}")  
            output = response['output']['message']['content'][0]['text'].strip()  
            score = float(output)  
            print(f"Score parsed: {score}")  
            return max(0.0, min(1.0, score))  
                  
        except Exception as e:  
            if "ThrottlingException" in str(e) and attempt < max_retries - 1:  
                time.sleep(2 ** attempt)  
            else:  
                print(f"Bedrock call failed: {e}")  
                return None  
    return None  
  
  
def lambda_handler(event, context):  
    """AWS Lambda handler - processes samples from RFTEvalInvoker."""  
    try:  
        samples = event if isinstance(event, list) else [event]  
        results = []  
          
        for sample in samples:  
            sample_id = sample.get("id", "unknown")  
            messages = sample.get("messages", [])  
              
            # Extract assistant response (response A)  
            response_a = ""  
            for msg in messages:  
                if msg.get("role") in ["assistant", "nova_assistant"]:  
                    response_a = msg.get("content", "")  
                    break  
              
            # Extract reference answer from root level (no longer in metadata)  
            reference_answer = sample.get("reference_answer", "")  
              
            # Handle both string and dict reference_answer formats  
            if isinstance(reference_answer, dict):  
                # If reference_answer is a dict, extract the explanation or compliant field  
                response_b = reference_answer.get("explanation", reference_answer.get("compliant", ""))  
            else:  
                response_b = reference_answer  
              
            if not response_a or not response_b:  
                results.append({  
                    "id": sample_id,  
                    "aggregate_reward_score": 0.0,  
                    "metrics_list": [{"name": "similarity_score", "value": 0.0, "type": "Metric"}]  
                })  
                continue  
              
            # Get similarity score  
            score = lambda_graded(response_a, response_b)  
              
            results.append({  
                "id": sample_id,  
                "aggregate_reward_score": score,  
                "metrics_list": [  
                    {  
                        "name": "similarity_score",  
                        "value": score,  
                        "type": "Metric"  
                    }  
                ]  
            })  
          
        return {"statusCode": 200, "body": json.dumps(results)}  
          
    except Exception as e:  
        print(f"Error: {e}")  
        return {"statusCode": 500, "body": json.dumps({"error": str(e)})}

Format input

Lambda menerima format input yang sama dengan fungsi reward lainnya:

{  
    "id": "sample-001",  
    "messages": [  
        {  
            "role": "user",  
            "content": "Do you have a dedicated security team?"  
        },  
        {  
            "role": "assistant",  
            "content": "As an AI developed by Amazon, I don't have a dedicated security team..."  
        }  
    ],  
    "reference_answer": {  
        "compliant": "No",  
        "explanation": "As an AI developed by Company, I do not have a traditional security team..."  
    },  
    "my_custom_field": "custom_value"  
}

Format output

{  
    "id": "sample-001",  
    "aggregate_reward_score": 0.85,  
    "metrics_list": [  
        {  
            "name": "similarity_score",  
            "value": 0.85,  
            "type": "Metric"  
        }  
    ]  
}

Pertimbangan deployment

Anda mungkin juga perlu menyesuaikan template prompt dan parameter inferensi berdasarkan kemampuan model dan format API yang Anda pilih.

Meningkatkan Throughput Batuan Dasar

Jika Anda mengalami pelambatan selama evaluasi, tingkatkan kuota model Bedrock Anda:

Logika coba ulang bawaan Lambda menangani pelambatan sesekali, tetapi evaluasi volume tinggi yang berkelanjutan memerlukan peningkatan kuota yang sesuai.

Kebijakan IAM yang Diperlukan:

{  
    "Version": "2012-10-17",		 	 	   
    "Statement": [  
        {  
            "Effect": "Allow",  
            "Action": [  
                "bedrock:InvokeModel"  
            ],  
            "Resource": "arn:aws:bedrock:*::foundation-model/*"  
        }  
    ]  
}