custom_llms

Creates, updates, deletes, gets or lists a custom_llms resource.

Overview

Name	custom_llms
Type	Resource
Id	databricks_workspace.agentbricks.custom_llms

Fields

The following fields are returned by SELECT queries:

get

Name	Datatype	Description
id	string
name	string
endpoint_name	string	Name of the endpoint that will be used to serve the custom LLM
agent_artifact_path	string
creation_time	string	Creation timestamp of the custom LLM
creator	string	Creator of the custom LLM
▶datasets	array	Datasets used for training and evaluating the model, not for inference
guidelines	array	Guidelines for the custom LLM to adhere to
instructions	string	Instructions for the custom LLM to follow
optimization_state	string	If optimization is kicked off, tracks the state of the custom LLM (CANCELLED, COMPLETED, CREATED, FAILED, PENDING, RUNNING)

Methods

The following methods are available for this resource:

Name	Accessible by	Required Params	Optional Params	Description
get	select	id, deployment_name		Get a Custom LLM.
create	insert	deployment_name, name, instructions		Create a Custom LLM.
update	update	id, deployment_name, custom_llm, update_mask		Update a Custom LLM.
delete	delete	id, deployment_name		Delete a Custom LLM.
cancel_optimize	exec	id, deployment_name		Cancel a Custom LLM Optimization Run.
start_optimize	exec	id, deployment_name		Start a Custom LLM Optimization Run.

Parameters

Parameters can be passed in the WHERE clause of a query. Check the Methods section to see which parameters are required or optional for each operation.

Name	Datatype	Description
deployment_name	string	The Databricks Workspace Deployment Name (default: dbc-abcd0123-a1bc)
id	string	The Id of the tile.

SELECT examples

get

Get a Custom LLM.

SELECT
id,
name,
endpoint_name,
agent_artifact_path,
creation_time,
creator,
datasets,
guidelines,
instructions,
optimization_state
FROM databricks_workspace.agentbricks.custom_llms
WHERE id = '{{ id }}' -- required
AND deployment_name = '{{ deployment_name }}' -- required
;

INSERT examples

create

Create a Custom LLM.

INSERT INTO databricks_workspace.agentbricks.custom_llms (
name,
instructions,
agent_artifact_path,
datasets,
guidelines,
deployment_name
)
SELECT 
'{{ name }}' /* required */,
'{{ instructions }}' /* required */,
'{{ agent_artifact_path }}',
'{{ datasets }}',
'{{ guidelines }}',
'{{ deployment_name }}'
RETURNING
id,
name,
endpoint_name,
agent_artifact_path,
creation_time,
creator,
datasets,
guidelines,
instructions,
optimization_state
;

Manifest

# Description fields are for documentation purposes
- name: custom_llms
props:
  - name: deployment_name
    value: "{{ deployment_name }}"
    description: Required parameter for the custom_llms resource.
  - name: name
    value: "{{ name }}"
    description: |
      Name of the custom LLM. Only alphanumeric characters and dashes allowed.
  - name: instructions
    value: "{{ instructions }}"
    description: |
      Instructions for the custom LLM to follow
  - name: agent_artifact_path
    value: "{{ agent_artifact_path }}"
    description: |
      This will soon be deprecated!! Optional: UC path for agent artifacts. If you are using a dataset that you only have read permissions, please provide a destination path where you have write permissions. Please provide this in catalog.schema format.
  - name: datasets
    description: |
      Datasets used for training and evaluating the model, not for inference. Currently, only 1 dataset is accepted.
    value:
      - table:
          table_path: "{{ table_path }}"
          request_col: "{{ request_col }}"
          response_col: "{{ response_col }}"
  - name: guidelines
    value:
      - "{{ guidelines }}"
    description: |
      Guidelines for the custom LLM to adhere to

UPDATE examples

update

Update a Custom LLM.

UPDATE databricks_workspace.agentbricks.custom_llms
SET 
custom_llm = '{{ custom_llm }}',
update_mask = '{{ update_mask }}'
WHERE 
id = '{{ id }}' --required
AND deployment_name = '{{ deployment_name }}' --required
AND custom_llm = '{{ custom_llm }}' --required
AND update_mask = '{{ update_mask }}' --required
RETURNING
id,
name,
endpoint_name,
agent_artifact_path,
creation_time,
creator,
datasets,
guidelines,
instructions,
optimization_state;

DELETE examples

delete

Delete a Custom LLM.

DELETE FROM databricks_workspace.agentbricks.custom_llms
WHERE id = '{{ id }}' --required
AND deployment_name = '{{ deployment_name }}' --required
;

Lifecycle Methods

cancel_optimize

Cancel a Custom LLM Optimization Run.

EXEC databricks_workspace.agentbricks.custom_llms.cancel_optimize 
@id='{{ id }}' --required, 
@deployment_name='{{ deployment_name }}' --required
;

start_optimize

Start a Custom LLM Optimization Run.

EXEC databricks_workspace.agentbricks.custom_llms.start_optimize 
@id='{{ id }}' --required, 
@deployment_name='{{ deployment_name }}' --required
;