Syntax Elements for AI_ImageEmbeddings - Teradata Analytics

Teradata® Text Analytics AI Functions

Product
Teradata Analytics
Release Number
20.00
Published
July 2026
ft:locale
en-US
ft:lastEdition
2026-07-10
dita:mapPath
foa1783708212383.ditamap
dita:ditavalPath
jlp1749501702948.ditaval
dita:id
foa1783708212383
Name Required or Optional Description Type Default Value Permitted Values
ON clause Required Specifies the table name, view name, or query as an InputTable. N/A N/A N/A
ImageColumn Required Specifies the name of the column with images to be used as input. string or binary N/A N/A
isDebug Optional Specifies whether error logging is required. boolean False True,False
Accumulate Optional Specifies the names of the input columns to copy to the output table. string N/A N/A
OutputFormat Optional Specifies the format for the resultant embeddings. string VARCHAR VECTOR, VECTOR32, VARCHAR, or VARBYTE
OutputPrecision Optional Specifies decimal precision for embedding values.

Applicable only to VARBYTE.

string DOUBLE FLOAT or DOUBLE
ShowNumTokens Optional Specifies whether to display token information in the query output. string 0 true, yes, 1, false, no, 0
RefreshCredentialTimeSeconds Optional Specifies how often to refresh credentials for AWS, Azure, and GCP. This applies when AWS, Azure, and GCP VMs connect to AWS Bedrock, Azure ML, and GCP Gemini models without needing external credentials. Integer 0 Valid double value and non-zero value to trigger refresh.