Skip to content

Latest commit

 

History

History
24 lines (14 loc) · 1.38 KB

UPGRADING.md

File metadata and controls

24 lines (14 loc) · 1.38 KB
Raw

Upgrading Concrete ML In Your Project

This guide aims to help developers who are upgrading from older versions of Concrete ML. Although we cannot cover everything, we have compiled the most important points to assist you in your transition. If you encounter any issues, please do not hesitate to reach out to https://community.zama.ai or ask on the fhe.org Discord.

Upgrading to 1.0.0

Please take note of the following changes when upgrading to version 1.0.0:

  • execute_in_fhe argument in .predict() methods has been replaced by fhe="disable|simulate|execute". The disable option runs the model in Python, while simulate performs a FHE simulation, and execute provides actual FHE execution.

  • encrypt_run_decrypt function can now only be executed in FHE. For FHE simulations, please use the simulate function instead.

  • In some models, the forward_fhe function has been renamed to fhe_circuit for consistency across all models.

  • verbose_compilation parameter has been renamed to verbose in the compile functions.

  • compilation_artifacts parameter has been renamed to artifacts.

  • use_virtual_lib parameter in concrete.ml.torch has been removed from the following functions:

    • compile_onnx_model
    • compile_torch_model
    • compile_brevitas_qat_model This means that models are now always converted to FHE, and only FHE-friendly models can be compiled.