6.9.3 Unload a Model

The ML_MODEL_UNLOAD routine unloads a model from MySQL HeatWave AutoML. For ML_MODEL_UNLOAD parameter descriptions, see ML_MODEL_UNLOAD.

Before You Begin

• Review the following

  • Train a Model

  • Load a Model

Unload a Model

If you are on MySQL 9.0.0 or higher, you can verify what models are currently loaded with the ML_MODEL_ACTIVE routine before and after unloading the model.

1. If you are on MySQL 9.0.0 or higher, verify what models are currently loaded with the ML_MODEL_ACTIVE routine.

   mysql> CALL sys.ML_MODEL_ACTIVE('all', @model_info);

2. Select the session variable created to view all loaded models.

   mysql> SELECT JSON_PRETTY(@model_info);
   +-----------------------------------------------------------+
   | JSON_PRETTY(@model_info)                                  |
   +-----------------------------------------------------------+
   | [
     {
       "total model size(bytes)": 50209
     },
     {
       "user1": [
         {
           "recommendation_use_case": {
             "format": "HWMLv2.0",
             "model_size(byte)": 15609
           }
         },
         {
           "recommendation_use_case2": {
             "format": "HWMLv2.0",
             "model_size(byte)": 8766
           }
         },
         {
           "recommendation_use_case3": {
             "format": "HWMLv2.0",
             "model_size(byte)": 8402
           }
         },
         {
           "recommendation_use_case4": {
             "format": "HWMLv2.0",
             "model_size(byte)": 17432
           }
         }
       ]
     }
   ] |
   +-----------------------------------------------------------+
   1 row in set (0.0411 sec)

3. Refer to the appropriate model handle to unload. Alternatively, use the session variable for the model handle.

   The following example unloads a model by using the model handle:

   mysql> CALL sys.ML_MODEL_UNLOAD('recommendation_use_case');

   Where:

   • recommendation_use_case is the model handle.

   The following example unloads a model by using the session variable for the model handle:

   mysql> CALL sys.ML_MODEL_UNLOAD(@recommendation_model);

   Where:

   • @recommendation_model is the assigned session variable for the model handle.

4. If you are on MySQL 9.0.0, run ML_MODEL_ACTIVE again to confirm the model is successfully unloaded

   mysql> CALL sys.ML_MODEL_ACTIVE('all', @model_info);
   mysql> SELECT JSON_PRETTY(@model_info);
   +-----------------------------------------------------------+
   | JSON_PRETTY(@model_info)                                  |
   +-----------------------------------------------------------+
   | [
     {
       "total model size(bytes)": 34600
     },
     {
       "user1": [
         {
           "recommendation_use_case2": {
             "format": "HWMLv2.0",
             "model_size(byte)": 8766
           }
         },
         {
           "recommendation_use_case3": {
             "format": "HWMLv2.0",
             "model_size(byte)": 8402
           }
         },
         {
           "recommendation_use_case4": {
             "format": "HWMLv2.0",
             "model_size(byte)": 17432
           }
         }
       ]
     }
   ] |
   +-----------------------------------------------------------+
   1 row in set (0.0411 sec)

   The list of loaded models shows the model is unloaded.

What's Next

• Review how to Create a Machine Learning Model.

• Review Machine Learning Use Cases to create machine learning models with sample datasets.