# Conversions

Learn how values move between numeric formats, layouts, and owners.

[View canonical HTML](https://pllm.run/sdk/build/conversions/)

Document ID: `pllm.docs.conversions`  
Release: `0.1.0`

Conversions connect components that use different formats. Examples include a
quantized tensor to ring values, arithmetic shares to bits, residues to signed
integers, a packed matrix to row-major layout, or server shares to client output.
Each conversion records bounds, rounding, network traffic, ownership, visible
information, and assurance requirements.

The compiler can insert only a declared conversion that satisfies both component
interfaces. A matching Python or Rust type cannot silently open a private value or
allow a lossy cast.

No public Python API currently exposes conversion planning or execution. See
[current Python support](/sdk/reference/status/). No built-in conversion component
is currently published, and that absence does not authorize an implicit cast.