mlos_bench.storage.sql.schema
DB schema definition for the SqlStorage
backend.
Notes
The SQL statements are generated by SQLAlchemy, but can be obtained using
repr
or str
(e.g., via print()
) on this object.
The mlos_bench
CLI will do this automatically if the logging level is set to
DEBUG
.
Also see the mlos_bench CLI usage for details on how to invoke only the schema creation/update routines.
Classes
A class to define and create the DB schema. |
Module Contents
- class mlos_bench.storage.sql.schema.DbSchema(engine: sqlalchemy.engine.Engine | None)[source]
A class to define and create the DB schema.
Declare the SQLAlchemy schema for the database.
- Parameters:
engine (sqlalchemy.engine.Engine | None) – The SQLAlchemy engine to use for the DB schema. Listed as optional for alembic schema migration purposes so we can reference it inside it’s
env.py
config file formeta
data inspection, but won’t generally be functional without one.
- __repr__() str [source]
Produce a string with all SQL statements required to create the schema from scratch in current SQL dialect.
That is, return a collection of CREATE TABLE statements and such. NOTE: this method is quite heavy! We use it only once at startup to log the schema, and if the logging level is set to DEBUG.
- Returns:
sql – A multi-line string with SQL statements to create the DB schema from scratch.
- Return type:
- update() DbSchema [source]
Updates the DB schema to the latest version.
Notes
Also see the mlos_bench CLI usage for details on how to invoke only the schema creation/update routines.
- Return type:
- config[source]
The Table storing
TunableConfigData
info.
- config_param[source]
The Table storing
TunableConfigData
info.
- experiment[source]
The Table storing
ExperimentData
info.
- property meta: sqlalchemy.MetaData[source]
Return the SQLAlchemy MetaData object.
- Return type:
sqlalchemy.MetaData
- objectives[source]
The Table storing
Experiment
optimization objectives info.