Direct Preference Optimization (DPO)

Direct Preference Optimization (DPO) ist eine effiziente Methode zur Optimierung von Basismodellen, bei der gepaarte Vergleichsdaten verwendet werden, um die Modellergebnisse an den menschlichen Präferenzen auszurichten. Dieser Ansatz ermöglicht eine direkte Optimierung des Modellverhaltens auf der Grundlage von menschlichem Feedback darüber, welche Reaktionen wünschenswerter sind.

Sowohl Full-Rank-DPO als auch Low-Rank Adapter (LoRA)-DPO sind verfügbar.

Anforderungen an das Datenformat

Sowohl bei Full-Rank- als auch bei LoRA-DPO ähneln die Anforderungen an das Trainingsdatenformat denen von SFT. Bei DPO muss der letzte Zug jedoch Präferenzpaare haben. Hier ein Beispiel für das DPO-Datenformat:

// N-1 turns same as SFT format
{
    "role": "assistant",
    "candidates": [
        {
            "content": [
                {
                    "text": "..."
                } // content list can contain multiple 'text' objects
            ],
            "preferenceLabel": "preferred"
        },
        {
            "content": [
                {
                    "text": "..."
                } // content list can contain multiple 'text' objects
            ],
            "preferenceLabel": "non-preferred"
        }
    ]
}

Hier ein weiteres vollständiges DPO-Textbeispiel:

{
    "system": [ 
        {
            "text": "..." 
        } 
    ],
    "messages":[
        {
            "role": "user",
            "content": [
                {
                    "text": "..."
                }
            ]
        },
        {
            "role": "assistant",
            "content": [
                {
                    "text": "..."
                }
            ]
        }, 
        {
            "role": "user",
            "content": [
                {
                    "text": "..."
                }
            ]
        },
        {
            "role": "assistant",
            "candidates": [
                {
                    "content": [
                        {
                            "text": "..."
                        }
                    ],
                    "preferenceLabel": "preferred"
                },
                {
                    "content": [
                        {
                            "text": "..."
                        }
                    ],
                    "preferenceLabel": "non-preferred"
                }
            ]
        }
    ],
}

Hier ein vollständiges DPO-Image-Beispiel:

{
    "system": [ 
        {
            "text": "..." 
        } 
    ],
    "messages":[
        {
            "role": "user",
            "content": [
                {
                    "text": "..."
                },
                {
                    "text": "..."
                },
                {
                    "image": {
                        "format": "jpeg",
                        "source": {
                            "s3Location": {
                                "uri": "s3://your-bucket/your-path/your-image.jpg",
                                "bucketOwner": "your-aws-account-id"
                            }
                        }
                    }
                } // "content" can have multiple "text" and "image" objects. 
                 // max image count is 10
            ]
        },
        {
            "role": "assistant",
            "content": [
                {
                    "text": "..."
                }
            ]
        },
        {
            "role": "user",
            "content": [
                {
                    "text": "..."
                },
                {
                    "text": "..."
                },
                {
                    "image": {
                        "format": "jpeg",
                        "source": {
                            "s3Location": {
                                "uri": "s3://your-bucket/your-path/your-image.jpg",
                                "bucketOwner": "your-aws-account-id"
                            }
                        }
                    }
                } // "content" can have multiple "text" and "image" objects. 
                 // max image count is 10
            ]
        },           
        {
            "role": "assistant",
            "candidates": [
                {
                    "content": [
                        {
                            "text": "..."
                        }
                    ],
                    "preferenceLabel": "preferred"
                },
                {
                    "content": [
                        {
                            "text": "..."
                        }
                    ],
                    "preferenceLabel": "non-preferred"
                }
            ]
        }
    ],
}

Es gelten weitere Einschränkungen für die Eingabedatensätze. Weitere Informationen finden Sie unter Datensatzeinschränkungen. Wir empfehlen, dass Sie mindestens 1 000 Präferenzpaare angeben, um ein effektives Training zu gewährleisten. Hochwertige Präferenzdaten führen zu effizienteren Ergebnissen.

Wir empfehlen, in den folgenden Szenarien DPO zu verwenden:

Full-Rank-DPO

Die Amazon-Nova-Parameter, die für Full-Rank-DPO verfügbar sind, lauten wie folgt:

Full-Rank-Rezept für DPO

Im Folgenden finden Sie ein Full-Rank-Rezept für DPO

## Run config
run:
  name: "my-dpo-micro-job"             # A descriptive name for your training job
  model_type: "amazon.nova-micro-v1:0:128k"  # Model variant specification, do not change
  model_name_or_path: "nova-micro/prod"      # Base model path, do not change
  replicas: 2                     # Number of compute instances for training, allowed values are 2, 4, 8
  data_s3_path: s3:Replace with your S3 bucket name/input.jsonl
  validation_data_s3_path: [OPTIONAL] s3:your S3 bucket name/input.jsonl
  output_s3_path: [S3_PATH_TO_STORE_MANIFEST]

## Training specific configs
training_config:
  max_length: 32768               # Maximum context window size (tokens).
  global_batch_size: 64           # Global batch size, allowed values are 16, 32, 64.

  trainer:
    max_epochs: 2                # Number of training epochs

  model:
    hidden_dropout: 0.0          # Dropout for hidden states, must be between 0.0 and 1.0
    attention_dropout: 0.0       # Dropout for attention weights, must be between 0.0 and 1.0
    ffn_dropout: 0.0             # Dropout for feed-forward networks, must be between 0.0 and 1.0

    optim:
      lr: 1e-5                 # Learning rate
      name: distributed_fused_adam  # Optimizer algorithm, do not change
      adam_w_mode: true        # Enable AdamW mode
      eps: 1e-06               # Epsilon for numerical stability
      weight_decay: 0.0        # L2 regularization strength, must be between 0.0 and 1.0
      betas:                   # Adam optimizer betas, must be between 0.0 and 1.0
        - 0.9
        - 0.999
      sched:
        warmup_steps: 10     # Learning rate warmup steps
        constant_steps: 0    # Steps at constant learning rate
        min_lr: 1e-6         # Minimum learning rate, must be lower than lr

    dpo_cfg:
        beta: 0.1               # Strength of preference enforcement. Limits: [0.001, 0.5]

    peft:
        peft_scheme: null        # Disable LoRA, trigger full rank fine tuning

Low-Rank Adapter DPO

Die Amazon-Nova-Parameter, die für Low-Rank Adapter DPO verfügbar sind, lauten wie folgt:

LoRA-DPO-Rezept

Das Folgende ist ein Rezept für LoRA DPO.

## Run config
run:
    name: "my-lora-run"             # A descriptive name for your training job
    model_type: "amazon.nova-lite-v1:0:300k"  # Model variant specification, do not change
    model_name_or_path: "nova-lite/prod"      # Base model path, do not change
    replicas: 4                     # Number of compute instances for training. All supported values: {4, 8, 16}
    data_s3_path: s3:Replace with your S3 bucket name/input.jsonl
    validation_data_s3_path: [OPTIONAL] s3:your S3 bucket name/input.jsonl
    output_s3_path: [S3_PATH_TO_STORE_MANIFEST]

## Training specific configs
training_config:
    max_length: 16384               # Maximum context window size (tokens). Should be between [1024, 32768] and multiple of 1024.
                                    # Note: Image dataset for DPO has a limit on 20k samples and 16384 max_length
    global_batch_size: 64           # Total samples per step. Limits: {16, 32, 64, 128, 256}

    trainer:
        max_epochs: 2               # Number of training epochs

    model:
        hidden_dropout: 0.0          # Dropout for hidden states. Limits: [0.0, 1.0]
        attention_dropout: 0.0       # Dropout for attention weights. Limits: [0.0, 1.0]
        ffn_dropout: 0.0             # Dropout for feed-forward networks. Limits: [0.0, 1.0]

        optim:
            lr: 1e-5                 # Learning rate
            name: distributed_fused_adam  # Optimizer algorithm, do not change
            adam_w_mode: true        # Enable AdamW mode
            eps: 1e-08               # Epsilon for numerical stability
            weight_decay: 0.01       # L2 regularization strength
            betas:                   # Adam optimizer betas. Limits: [0.0, 1.0]
                - 0.9
                - 0.999
            sched:
                warmup_steps: 10     # Learning rate warmup steps
                constant_steps: 0    # Steps at constant learning rate
                min_lr: 1e-6         # Minimum learning rate

        dpo_cfg:
            beta: 0.01               # Strength of preference enforcement. Limits: [0.001, 0.5]

        peft:
            peft_scheme: "lora"      # Enable LoRA for parameter-efficient fine-tuning
            lora_tuning:
                loraplus_lr_ratio: 20.0  # LoRA+ learning rate scaling factor. Limits: [0.0, 100.0]
                alpha: 64            # Scaling factor for LoRA weights. [32, 64, 96, 128, 160, 192]
                adapter_dropout: 0.01  # Regularization for LoRA parameters. Limits: [0.0, 1.0]

Einschränkungen

Für DPO gelten folgende Einschränkungen: