Ctrl+K
Logo image Logo image

Site Navigation

  • API Reference
  • Examples

Site Navigation

  • API Reference
  • Examples

Section Navigation

  • PyFlink Table
  • PyFlink DataFrame
    • DataFrame
    • DataFrame Creation
    • Input/Output
    • SQL
    • Data Types
    • User Defined Functions
    • Configuration
    • Catalog
    • GPU Support
    • AI / LLM
  • PyFlink Multimodal
  • PyFlink DataStream
  • PyFlink Common

pyflink.dataframe.Provider#

class Provider[source]#

Base class for model providers.

A provider owns the Python-side configuration for a Java model provider factory. Subclasses translate Python-style constructor parameters to the Java-side option keys expected by Flink’s ModelDescriptor.

to_options() is called in two modes:

  • Without input_columns, it serializes the provider’s static configuration.

  • With input_columns, it may additionally validate or adapt options against the model input schema before the descriptor is created.

Custom provider subclasses should accept the optional input_columns argument even if they do not need schema-aware option handling.

Methods

model_option_key()

Return the Java-side option key used for a per-call model name.

provider_name()

Return the provider identifier recognized by Flink's Java runtime.

to_options([input_columns])

Return Java-side options for ModelDescriptor.

previous

pyflink.dataframe.list_providers

next

pyflink.dataframe.OpenAICompatProvider

Show Source

Created using Sphinx 4.5.0.