Skip to content

/AWS1/CL_SGMMDELSPECULATIVED00

Settings for the model speculative decoding technique that's applied by a model optimization job.

CONSTRUCTOR

IMPORTING

Required arguments:

iv_technique TYPE /AWS1/SGMMDELSPECULATIVEDECO00 /AWS1/SGMMDELSPECULATIVEDECO00

The speculative decoding technique to apply during model optimization.

Optional arguments:

io_trainingdatasource TYPE REF TO /AWS1/CL_SGMMDELSPECULATIVED01 /AWS1/CL_SGMMDELSPECULATIVED01

The location of the training data to use for speculative decoding. The data must be formatted as ShareGPT, OpenAI Completions or OpenAI Chat Completions. The input can also be unencrypted captured data from a SageMaker endpoint as long as the endpoint uses one of the above formats.


Queryable Attributes

Technique

The speculative decoding technique to apply during model optimization.

Accessible with the following methods

Method Description
GET_TECHNIQUE() Getter for TECHNIQUE, with configurable default
ASK_TECHNIQUE() Getter for TECHNIQUE w/ exceptions if field has no value
HAS_TECHNIQUE() Determine if TECHNIQUE has a value

TrainingDataSource

The location of the training data to use for speculative decoding. The data must be formatted as ShareGPT, OpenAI Completions or OpenAI Chat Completions. The input can also be unencrypted captured data from a SageMaker endpoint as long as the endpoint uses one of the above formats.

Accessible with the following methods

Method Description
GET_TRAININGDATASOURCE() Getter for TRAININGDATASOURCE