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 class

ModelConfig

ModelConfig

class max.pipelines.kv_cache.ModelConfig(*args, **kwargs)

source

Bases: Protocol

Structural protocol for model configuration consumed by MemoryPlanner.

Any object that exposes devices satisfies this protocol. Planners that also need KV-cache parameters check for get_kv_params via isinstance or guard it with hasattr.

devices

property devices: list[Device]

source

Returns the list of devices on which the model runs.