Answer Conversion

A guide for bidirectional V1/V2 conversion of answer annotations.

Data Structure

V1 Structure

# annotations
{
    "id": "answer_1",
    "tool": "answer",
    "isLocked": False,
    "isVisible": True,
    "isValid": True,
    "classification": {
        "class": "response"
    },
    "label": ["response"]
}

# annotationsData
{
    "id": "answer_1",
    "tool": "answer",
    "model": "gpt-4",
    "output": [
        {
            "type": "text",
            "value": "This image shows a cat sitting on a couch.",
            "primaryKey": "main",
            "changeHistory": []
        }
    ],
    "displayName": "GPT-4 Response",
    "generatedBy": "ai",
    "promptAnnotationId": "PromptAbc123"  # linked prompt
}

V2 Structure

{
    "id": "answer_1",
    "classification": "response",
    "attrs": [],
    "data": {
        "model": "gpt-4",
        "output": [
            {
                "type": "text",
                "value": "This image shows a cat sitting on a couch.",
                "primaryKey": "main",
                "changeHistory": []
            }
        ],
        "displayName": "GPT-4 Response",
        "generatedBy": "ai",
        "promptAnnotationId": "PromptAbc123"
    }
}

Conversion Rules

V1 → V2

V1 Field	V2 Field
output	data.output
model	data.model
displayName	data.displayName
generatedBy	data.generatedBy
promptAnnotationId	data.promptAnnotationId
classification.class	classification
classification.{other}	attrs[{name, value}]

V2 → V1

V2 Field	V1 Field
data.output	output
data.model	model
data.displayName	displayName
data.generatedBy	generatedBy
data.promptAnnotationId	promptAnnotationId
classification	classification.class
attrs[{name, value}]	classification.{name: value}

Usage Examples

Basic Conversion

from synapse_sdk.utils.converters.dm import convert_v1_to_v2, convert_v2_to_v1

# V1 answer data
v1_data = {
    "annotations": {
        "prompt_1": [
            {
                "id": "AnswerXyz789",
                "tool": "answer",
                "classification": {"class": "response"}
            }
        ]
    },
    "annotationsData": {
        "prompt_1": [
            {
                "id": "AnswerXyz789",
                "tool": "answer",
                "model": "gpt-4",
                "output": [
                    {
                        "type": "text",
                        "value": "This image shows a cat sitting on a couch.",
                        "primaryKey": "main",
                        "changeHistory": []
                    }
                ],
                "displayName": "GPT-4 Response",
                "generatedBy": "ai",
                "promptAnnotationId": "PromptAbc123"
            }
        ]
    }
}

# Convert to V2
result = convert_v1_to_v2(v1_data)
annotation_data = result["annotation_data"]

# Check V2 result
answer = annotation_data["prompts"][0]["answer"][0]
print(answer["data"]["output"])  # [{"type": "text", "value": "...", ...}]
print(answer["data"]["model"])  # "gpt-4"
print(answer["data"]["promptAnnotationId"])  # "PromptAbc123"

Prompt-Answer Pair Processing

# Data with both prompt and answer
v1_qa = {
    "annotations": {
        "prompt_1": [
            {"id": "Q1", "tool": "prompt", "classification": {"class": "question"}},
            {"id": "A1", "tool": "answer", "classification": {"class": "response"}}
        ]
    },
    "annotationsData": {
        "prompt_1": [
            {
                "id": "Q1",
                "tool": "prompt",
                "input": [{"type": "text", "value": "What color is the cat?"}]
            },
            {
                "id": "A1",
                "tool": "answer",
                "model": "human",
                "output": [{"type": "text", "value": "The cat is orange."}],
                "generatedBy": "human",
                "promptAnnotationId": "Q1"
            }
        ]
    }
}

# Convert to V2
result = convert_v1_to_v2(v1_qa)

# Both prompt and answer included
prompts = result["annotation_data"]["prompts"][0].get("prompt", [])
answers = result["annotation_data"]["prompts"][0].get("answer", [])

print(f"Prompts: {len(prompts)}, Answers: {len(answers)}")

Roundtrip Verification

def verify_answer_roundtrip(v1_original):
    """Verify answer roundtrip"""
    # V1 → V2 → V1
    v2_result = convert_v1_to_v2(v1_original)
    v1_restored = convert_v2_to_v1(v2_result)

    # Compare output
    orig_data = v1_original["annotationsData"]["prompt_1"][0]
    rest_data = v1_restored["annotationsData"]["prompt_1"][0]

    assert orig_data["output"] == rest_data["output"]
    assert orig_data["model"] == rest_data["model"]
    assert orig_data["promptAnnotationId"] == rest_data["promptAnnotationId"]

    print("Answer roundtrip verification successful")

verify_answer_roundtrip(v1_data)

Field Descriptions

Field	Description
output	Answer content array
model	Generation model (e.g., gpt-4, human)
displayName	Display name
generatedBy	Generator (ai/human)
promptAnnotationId	Linked prompt ID

Related Documentation

• Prompt Conversion - Prompt annotations