27.3.9.8 Convenience Methods

The GenAI API provides the convenience methods described in this section under the ml namespace. These methods act as wrappers for the LLM methods; rather than being invoked as LLM instance methods, they take the model ID as one of the options passed to them. ml.generate() and ml.rag() return only the text portions of the values returned by their LLM counterparts.

ml.generate()

This method is a wrapper for LLM.generate(). It loads the model specified by the model_id specified as one of the options, generates a response based on the prompt using this model, and returns the response. The default model_id ("cohere.command") is used if one is not specified. Like the LLM method, ml.generate() supports two variants, one for a single invocation, and one for batch processing.

Signature (single job)

String ml.generate(
  String prompt,
  Object options
)

Arguments

prompt (String) (default "cohere.command"): The desired prompt
options (Object) (default {}): The options employed for generation; these follow the same rules as the options used with LLM.generate()

Return type

String: The text of the response

Usage

//  Both invocations use "cohere.command" as the model_id

let response = ml.generate("What is Mysql?", {max_tokens: 10})

let response = ml.generate("What is Mysql?", {model_id: "cohere.command", max_tokens: 10})

Signature (batch processing)

undefined ml.generate(
  Table inputTable, 
  String inputColumn, 
  String outputColumn, 
  Object options
)

Arguments

inputTable (Table): Table to use for operations
inputColumn (String): Name of column from inputTable to be embedded
outputColumn (String): Name of column in which to store embeddings; this can be either a fully-qualified name of a column or the name of the column only; in the latter case, the input table and its schema are used to construct the fully-qualified name
options (Object) (optional; default {}): An object containing the options used for embedding; see the description of ML_EMBED_ROW for available options

Return type

undefined

Usage

let schema = session.getSchema("mlcorpus")
let table = schema.getTable("genai_table")

ml.generate(table, "input", "mlcorpus.predictions.response", {max_tokens: 10})

ml.embed()

This method is a wrapper for LLM.embed(). Like the LLM method, it supports two variants, one for a single invocation, and one for batch processing.

Signature (single job)

Float32Array ml.embed(
  String query,
  Object options
)

Arguments

query (String): Text of a natural-language query
options (Object) (default {}): The options employed for generation; these follow the same rules as the options used with LLM.embed(); the default model_id is "all_minilm_l12_v2"

Return type

Float32Array (MySQL VECTOR): The embedding

Usage

//  These produce the same result

let embedding = ml.embed("What is Mysql?", {model_id: "all_minilm_l12_v2"})

let embedding = ml.embed("What is Mysql?", {})

Signature (batch processing)

undefined ml.embed(
  Table inputTable,
  String inputColumn,
  String outputColumn,
  Object options
)

Arguments

inputTable (Table): Table to use for operations
inputColumn (String): Name of column from inputTable to be embedded
outputColumn (String): Name of column in which to store embeddings; this can be either a fully-qualified name of a column or the name of the column only; in the latter case, the input table and its schema are used to construct the fully-qualified name
options (Object) (optional; default {}): An object containing the options used for embedding; see the description of ML_EMBED_ROW for available options

Return type

undefined

Usage

let schema = session.getSchema("mlcorpus")
let table = schema.getTable("genai_table")

ml.embed(table, "input", "mlcorpus.predictions.response", 
        {model_id: "all_minilm_l12_v2"})

ml.load()

This static method loads an existing (and already trained) HeatWave AutoML model having the name specified. An error is thrown if model with the given name does not exist.

Signature

```
Object ml.load(
  String name
)
```

Arguments

name (String): The name of the model.

Return type

Object: Any of Classifier, Regressor, Forecaster, AnomalyDetector, or Recommender, depending on the type of model loaded.

For more information, see ML_MODEL_LOAD.

ml.rag()

This is a wrapper for LLM.rag(). Like the LLM method, it supports two variants, one for a single invocation, and one for batch processing.

Signature (single job)

String ml.rag(
  String query,
  Object options
)

Arguments

query (String): Text of a natural-language query
options (Object) (default {}): The options employed for generation; these follow the same rules as the options used with LLM.rag(); the default model_id is "mistral-7b-instruct-v1"

Return type

String: Response text

Usage

//  These produce the same result

let result = ml.rag("What is MySql?", {schema: ["vector_store"], n_citations: 1, model_options: {model_id: "mistral-7b-instruct-v1"}})

let result = ml.rag("What is MySql?", {schema: ["vector_store"], n_citations: 1})

Signature (batch processing)

undefined ml.rag(
  Table inputTable,
  String inputColumn,
  String outputColumn,
  Object options
)

Arguments

inputTable (Table): Table to use for operations
inputColumn (String): Name of column from inputTable to be embedded
outputColumn (String): Name of column in which to store embeddings; this can be either a fully-qualified name of a column or the name of the column only; in the latter case, the input table and its schema are used to construct the fully-qualified name
options (Object) (optional; default {}): An object containing the options used for embedding; see the description of ML_EMBED_ROW for available options

Return type

undefined

Usage

let schema = session.getSchema("mlcorpus")
let table = schema.getTable("genai_table")

ml.rag(table, "input", "mlcorpus.predictions.response", 
      {schema: ["vector_store"], n_citations: 1, model_options: 
      {model_id: "mistral-7b-instruct-v1"}});