Problem with a custom datapoint: slow to be created

[LStromann]

=Hello guys, I am trying to make my knowledge graph with a custom model, but it is so slow (I am using gpt-5-mini, and it is almost impossible to create, all the time on the task "extract_graph_data"). Does anyone have any tips to solve this? Before i tried a custom task, but continued slowly. Using NEO4J and GPT-5

from typing import List
from cognee.infrastructure.engine import DataPoint


class Entity(DataPoint):
    """
    Node of an entity, which can be an API, Endpoint, Service, etc.
    """

    title: str
    description: str
    metadata: dict = {
        "index_fields": [
            "title",
        ]
    }


class Example(DataPoint):
    "An example from a rule or problem"

    title: str
    description: str
    entities: List[Entity]
    metadata: dict = {"index_fields": ["title", "description"]}
    wiki_url: List[str]


class Solution(DataPoint):
    "A solution to a problem"

    title: str
    description: str
    effectiveness: str
    entities: List[Entity]
    metadata: dict = {"index_fields": ["title", "description"]}
    wiki_url: List[str]


class Problem(DataPoint):
    "A problem"

    title: str
    description: str
    status: str
    priority: str
    url: List[str]
    examples: List[Example]
    solutions: List[Solution]
    entities: List[Entity]
    metadata: dict = {"index_fields": ["title", "description"]}


class Rule(DataPoint):
    "A rule"

    title: str
    examples: List[Example]
    description: str
    entities: List[Entity]
    metadata: dict = {"index_fields": ["title", "description"]}
    url: List[str]


class KnowledgeGraph(DataPoint):
    problems: List[Problem]
    rules: List[Rules] 

This discussion was automatically pulled from Discord.

[LStromann]

I found a way, set fields as optional :), will be very fast