For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python function

registered_pipeline_model

registered_pipeline_model()

max.pipelines.lib.registration_graph_api.registered_pipeline_model(task, config, modules, batching)

source

Returns the pipeline model that serves a module registration.

The task picks the pipeline model, since each task prepares its inputs differently; every registration of a task shares it. An embeddings-generation registration gets encoder_pipeline_model()’s. A text-generation registration whose DECODER is a graph-API module gets graph_api_pipeline_model()’s; any other gets module_pipeline_model()’s.

Raises:

NotImplementedError – If the task is neither text nor embeddings generation, which is all registered modules serve so far.

Parameters:

• task (PipelineTask)
• config (type[ArchConfig])
• modules (tuple[tuple[str, type[Any]], ...])
• batching (type[Any] | None)

Return type:

type[GraphApiDecoderPipelineModel | Any]