afe.tvm_converter.parameters

Global parameters to the TVM converter. This holds data that is passed to the converter and is globally available during the TVM converter.

Classes

TVMConverterParams

Parameters to an invocation of the TVM converter. These parameters

Module Contents

class afe.tvm_converter.parameters.TVMConverterParams[source]

Parameters to an invocation of the TVM converter. These parameters control how to convert a Relay IR module to an AwesomeNet.

Parameters:

  • requantization_mode – Requantization mode to use when translating quantized operators.

  • is_quantized – Whether the module is already quantized. If true, then the translation will not prepare for the SiMa IR quantization pass to run afterward.

  • target – A target platform that a model is compiled for.

requantization_mode: afe.ir.defines.RequantizationMode[source]
is_quantized: bool = False[source]
target: sima_utils.common.Platform[source]