Configuring additional models in watsonx BI

You can change your model configuration to change the way that watsonx BI handles certain workloads.

Before you begin

The watsonx BI service must be installed. For more information about the resources that are required, see GPU requirements for models.

About this task

When you install watsonx BI, OpenAI gpt-oss-120b, granite-4-h-small, and ibm-slate-30m-english-rtrvr are automatically deployed. If you have heavy workloads, you can choose to run the ibm-slate-30m-english-rtrvr model on GPU.

Procedure

Complete the following steps to install the ibm-slate-30m-english-rtrvr model on GPU:

  1. Check which models are installed:
    oc get watsonxaiifm watsonxaiifm-cr \
-n ${PROJECT_CPD_INST_OPERANDS} \
-o jsonpath="{.spec.install_model_list}"
  2. Run the ibm-slate-30m-english-rtrvr model on GPU by modifying the custom resource:
    oc patch watsonxaiifm watsonxaiifm-cr \
--namespace=${PROJECT_CPD_INST_OPERANDS} \
--type=merge \
--patch='{"spec":{"ibm_slate_30m_english_rtrvr_replicas": 1, "ibm_slate_30m_english_rtrvr_resources": {"limits": {"cpu": "2", "memory": "4Gi", "nvidia.com/gpu": 1}, "requests": {"cpu": "1", "memory": "4Gi", "nvidia.com/gpu": 1}}}}'
  3. Update the watsonxaiifm custom resource:
    oc patch watsonxaiifm-cr \
-n ${PROJECT_CPD_INST_OPERANDS} \
-\-type=merge \
-\-patch="{\"spec\":{\"installed_model_list\":[${MODEL_LIST}]}}"
  4. Confirm that the model was added to the spec section of the watsonxaiifm custom resource:
    oc get watsonxaiifm watsonxaiifm-cr \
-n ${PROJECT_CPD_INST_OPERANDS} \
-o jsonpath="{.spec.install_model_list}"

What to do next

watsonx BI is ready to use. To begin using watsonx BI, see Getting started with watsonx BI.