ReasonersConfig

relationalai.config.config_reasoners_fields

Configure defaults for the reasoners API.

This section lives under reasoners in raiconfig.yaml and controls the polling behavior for long-running operations, legacy Direct Access base URL compatibility, and default settings for each reasoner type (logic, predictive, prescriptive). Whether the SDK uses Direct Access is controlled by the top-level direct_access flag on Config.

This section covers config-managed reasoner defaults only. Operational lifecycle settings such as auto_suspend_mins and await_storage_vacuum are managed outside config through the reasoner-management APIs or CLI.

Examples

Minimal YAML (in raiconfig.yaml):

default_connection: sf
connections:
    sf:
        type: snowflake
        # ...
reasoners:
    poll_max_delay_s: 2.0
    logic:
        size: HIGHMEM_X64_S
        query_timeout_mins: 30

Configure reasoners settings using a dict:

>>> from relationalai.config import create_config
>>> cfg = create_config(reasoners={
...     "poll_max_delay_s": 2.0,
...     "logic": {"size": "HIGHMEM_X64_S", "query_timeout_mins": 30},
... })

Configure reasoners settings using an explicit ReasonersConfig instance:

>>> from relationalai.config import create_config, ReasonersConfig, LogicReasonerConfig
>>> cfg = create_config(
...     reasoners=ReasonersConfig(
...         poll_max_delay_s=2.0,
...         logic=LogicReasonerConfig(size="HIGHMEM_X64_S", query_timeout_mins=30),
...     ),
... )

Attributes

.direct_access_base_url

ReasonersConfig.direct_access_base_url: (str, optional)

Legacy compatibility field for Direct Access base URL. Newer DA wiring resolves service endpoints dynamically.

.show_all_sizes

ReasonersConfig.show_all_sizes: bool

Show all available reasoner sizes (includes internal/dev sizes).

.poll_initial_delay_s

ReasonersConfig.poll_initial_delay_s: float

Initial delay between polls when waiting for operations to complete.

.poll_overhead_rate

ReasonersConfig.poll_overhead_rate: float

Exponential backoff growth rate per poll (for example, 0.2 means +20% per poll).

.poll_max_delay_s

ReasonersConfig.poll_max_delay_s: float

Maximum poll delay.

.logic

ReasonersConfig.logic: LogicReasonerConfig

Defaults for the logic reasoner.

.predictive

ReasonersConfig.predictive: ReasonerConfig

Defaults for the predictive reasoner.

.prescriptive

ReasonersConfig.prescriptive: ReasonerConfig

Defaults for the prescriptive reasoner.

Inheritance Hierarchy

Referenced By

RelationalAI Documentation
└──  Build With RelationalAI
    └──  Understand how PyRel works > PyRel configuration overview
        └──  Configure reasoners
            └──  Configure reasoner polling behavior