Dnn.Quantization.Weight.PrecisionPlan Module Reference

Concepts
concept	Mila::Dnn::Quant::Weight::AttentionPrecisionRoles
	Fused QKV projection and the attention output projection.
concept	Mila::Dnn::Quant::Weight::DecoderPrecisionPlan
	What an attention-plus-feed-forward decoder block requires.
concept	Mila::Dnn::Quant::Weight::EmbeddingPrecisionRole
	The embedding table, required at the network level rather than the block.
concept	Mila::Dnn::Quant::Weight::FeedForwardPrecisionRoles
	Fused gate+up projection and the down projection of a gated feed-forward network.
concept	Mila::Dnn::Quant::Weight::LanguageModelHeadRole
	The output (language-model head) table.

Classes
struct	Mila::Dnn::Quant::Weight::UniformPrecisionPlan< TPolicy >
	Assigns one policy to every role. More...

Typedefs
template<typename T>
using	Mila::Dnn::Quant::Weight::PrecisionPlanFor = typename Detail::PrecisionPlanSelector<T, WeightQuantPolicy<T>>::type
	Accepts either a plan or a bare policy, and yields a plan.

Files
file	Mila/Src/Dnn/Quantization/Weight/PrecisionPlan.ixx