pllm.preparation Python API

Preparation components define offline inventory production and trusted preparation roles without placing preparation work on the online inference path. Presence documents API availability, not complete model, security, or deployment coverage; check implementation status before relying on a path.

Research context

Relevant design inputs: DASH, ReDASH. PLLM's implementation and evidence claims remain independent.

Python SDK example

The example performs a bounded offline task and checks an observable result. It does not download a checkpoint or contact a provider.

import pllm.preparation as preparation

provider = preparation.ModelAwareCorrections()
assert provider.component == "pllm/model-aware-corrections"

API: pllm.preparation

Objects and signatures

BFVCorrelations

Selects component pllm/bfv-correlations/v1 during preparation; it provides bfv-correlation-generation for roles client, preparation.

• Canonical object: pllm.preparation.BFVCorrelations
• Kind: class
• Signature/type: __init__(self, *, poly_modulus_degree: int=4096) -> None
• Public exports: pllm.preparation.BFVCorrelations
• Public members: descriptor: ComponentDescriptor; __init__(self, *, poly_modulus_degree: int=4096) -> None; classmethod describe(cls) -> ComponentDescriptor

HEAuthenticatedPreprocessing

Selects component pllm/he-authenticated-preprocessing during preparation; it provides he-authenticated-preprocessing for roles client, preparation.

• Canonical object: pllm.preparation.HEAuthenticatedPreprocessing
• Kind: class
• Signature/type: __init__(self, *, poly_modulus_degree: int=4096, threads: int=1, enable_linear_correlations: bool=True) -> None
• Public exports: pllm.preparation.HEAuthenticatedPreprocessing
• Public members: descriptor: ComponentDescriptor; __init__(self, *, poly_modulus_degree: int=4096, threads: int=1, enable_linear_correlations: bool=True) -> None; classmethod describe(cls) -> ComponentDescriptor

ModelAwareCorrections

Selects component pllm/model-aware-corrections during preparation; it provides one-use-masked-linear-corrections for roles preparation.

• Canonical object: pllm.preparation.ModelAwareCorrections
• Kind: class
• Signature/type: __init__(self) -> None
• Public exports: pllm.ModelAwareCorrections, pllm.preparation.ModelAwareCorrections
• Public members: descriptor: ComponentDescriptor; __init__(self) -> None; classmethod describe(cls) -> ComponentDescriptor

PreparationProvider

Immutable PreparationProvider record carrying component, params for the workflow where preparation components define offline inventory production and trusted preparation roles without placing preparation work on the online inference path.

• Canonical object: pllm.preparation.PreparationProvider
• Kind: class
• Signature/type: constructor not exposed
• Public exports: pllm.preparation.PreparationProvider

create_preparation_app

Builds the preparation ASGI application from validated runtime dependencies.

• Canonical object: pllm.runtime.preparation_server.create_preparation_app
• Kind: function
• Signature/type: (config: 'GatewayConfig', engine: 'MaskedTransformerEngine', *, audit_hook: 'Callable[[str, bytes], None] \| None' = None, push_client: 'httpx.AsyncClient \| None' = None) -> 'FastAPI'
• Public exports: pllm.create_preparation_app, pllm.preparation.create_preparation_app