# ModelSpeculativeDecodingConfig
<a name="API_ModelSpeculativeDecodingConfig"></a>

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

## Contents
<a name="API_ModelSpeculativeDecodingConfig_Contents"></a>

 ** Technique **   <a name="sagemaker-Type-ModelSpeculativeDecodingConfig-Technique"></a>
The speculative decoding technique to apply during model optimization.  
Type: String  
Valid Values: `EAGLE`   
Required: Yes

 ** TrainingDataSource **   <a name="sagemaker-Type-ModelSpeculativeDecodingConfig-TrainingDataSource"></a>
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.  
Type: [ModelSpeculativeDecodingTrainingDataSource](API_ModelSpeculativeDecodingTrainingDataSource.md) object  
Required: No

## See Also
<a name="API_ModelSpeculativeDecodingConfig_SeeAlso"></a>

For more information about using this API in one of the language-specific Amazon SDKs, see the following:
+  [Amazon SDK for C\+\+](https://docs.amazonaws.cn/goto/SdkForCpp/sagemaker-2017-07-24/ModelSpeculativeDecodingConfig) 
+  [Amazon SDK for Java V2](https://docs.amazonaws.cn/goto/SdkForJavaV2/sagemaker-2017-07-24/ModelSpeculativeDecodingConfig) 
+  [Amazon SDK for Ruby V3](https://docs.amazonaws.cn/goto/SdkForRubyV3/sagemaker-2017-07-24/ModelSpeculativeDecodingConfig)