Mila
Deep Neural Network Library
Loading...
Searching...
No Matches
Mila::Dnn::Quant::Weight::UniformPrecisionPlan< TPolicy > Struct Template Referenceexport

Assigns one policy to every role. More...

Inheritance diagram for Mila::Dnn::Quant::Weight::UniformPrecisionPlan< TPolicy >:
Mila::Dnn::QwenUniformPrecisionPlan< NoWeightQuant > Mila::Dnn::QwenUniformPrecisionPlan< T > Mila::Dnn::QwenUniformPrecisionPlan< TPolicy >

Public Types

using EmbeddingTable = TPolicy
using FeedForwardDown = TPolicy
using FeedForwardGateUp = TPolicy
using LanguageModelHead = TPolicy
using OutputProjection = TPolicy
using QkvProjection = TPolicy

Detailed Description

template<WeightQuantPolicy TPolicy>
struct Mila::Dnn::Quant::Weight::UniformPrecisionPlan< TPolicy >

Assigns one policy to every role.

The bridge that keeps a bare policy a valid argument wherever a plan is expected, so families with no per-role need are untouched by this header existing. A role added here later must be added to this struct too – that is the definition of uniform, not an oversight.


The documentation for this struct was generated from the following file:
  • Mila/Src/Dnn/Quantization/Weight/PrecisionPlan.ixx