save embedding

The save embedding command saves trained model embeddings and their vocabulary to a vector table. The saved weights may be used as pretrained embeddings while training a downstream model or may be defined as dataset features from the vector database source using functions such as top-k nearest.

Syntax

save embedding <model name>,<run id>,<vector table name> 

Parameters

Parameter	Description
<model name>	The name of the ML model
<run id>	The specific training run identifier
<vector table name>	The vector database table name in which to store the embeddings and vocabulary

Example

save embedding trip_fare_2_ml_model,1,faredata