Python model configurations
Use dbt.config() to set configs for a Python model. This page lists the parameters needed. For how to write a Python model, refer to Python models.
dbt.config() only accepts literal values (strings, booleans, and numbers). For anything more complex, set the config in a properties YAML file. Refer to Configuring Python models.
Common parameters
The following parameters work on every platform that supports Python models.
| Parameter | Type | Description |
|---|---|---|
materialized | <string> | How dbt builds the model, such as table or incremental. Python models don't support view. |
tags | <list> | Tags to apply to the model. |
meta | <dictionary> | Custom metadata. Read it with dbt.config.meta_get(). |
packages | <list> | Packages to install, with optional version pins such as numpy==1.23.1. Required on some platforms. Refer to Configuring packages. |
Read any config back inside your model with dbt.config.get().
Platform-specific parameters
- Snowflake
- Databricks
- BigQuery
| Parameter | Type | Default | Description |
|---|---|---|---|
python_version | <string> | 3.10 | Python version the Snowpark model runs on. |
packages | <list> | [] | Anaconda packages to install when the model runs. dbt always adds snowflake-snowpark-python. |
imports | <list> | [] | Staged files, such as zip files, to make available to your model. Refer to Third-party Snowflake packages. |
external_access_integrations | <list> | [] | Snowflake external access integrations that let your model reach external APIs. |
secrets | <dictionary> | {} | Maps a variable name in your code to a Snowflake secret. Use it with external_access_integrations. |
use_anonymous_sproc | <boolean> | True | When True, dbt runs your model as an anonymous stored procedure. When False, dbt creates a named procedure (<model_name>__dbt_sp), calls it, then drops it. |
For more details and a more comprehensive list, refer to Snowflake configurations.
| Parameter | Type | Default | Description |
|---|---|---|---|
submission_method | <string> | all_purpose_cluster | How dbt submits the job: all_purpose_cluster, job_cluster, serverless_cluster, or workflow_job. |
create_notebook | <boolean> | false | With all_purpose_cluster, upload a notebook and run it as a job instead of using the Command API. |
cluster_id | <string> | None | ID of an existing all-purpose cluster to run on. |
http_path | <string> | None | Path of an existing all-purpose cluster to run on. |
job_cluster_config | <dictionary> | {} | Settings for the new cluster used by job_cluster and workflow_job. |
timeout | <integer> | 0 (no timeout) | Maximum time to wait for the command or job to run. |
packages | <list> | [] | Packages to install on the cluster. |
index_url | <string> | None (uses PyPI) | URL to install packages from. |
additional_libs | <list> | [] | Libraries to attach to the job. |
access_control_list | <list> | none | Access control for the job. |
python_job_config | <dictionary> | {} | Extra job and workflow settings. |
Not every parameter works with every submission method. For more details and a more comprehensive list, refer to Databricks configurations.
| Parameter | Type | Default | Description |
|---|---|---|---|
submission_method | <string> | none | Where the job runs: serverless, bigframes, or cluster. |
dataproc_cluster_name | <string> | none | Dataproc cluster to run on. Required for cluster. |
dataproc_region | <string> | none | Google Cloud region where Dataproc is enabled. |
gcs_bucket | <string> | none | Cloud Storage bucket for job artifacts. |
compute_region | <string> | none | Region of the job. |
notebook_template_id | <integer> | none | Colab Enterprise runtime template. |
packages | <list> | none | Packages to pip install. bigframes only. |
timeout | <integer> | none | Maximum run time in seconds. |
enable_list_inference | <boolean> | True | Let PySpark read multiple records in one operation. |
intermediate_format | <string> | parquet | File format for intermediate writes: parquet or orc. |
For more details and a more comprehensive list, refer to BigQuery configurations.
Was this page helpful?
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
