Optimization and training dynamics · MLD-0509

He initialization

Weight scaling designed for rectifier-like activations using fan-based variance.

Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.

Original example

Initialize a ReLU network.

Original caveat

Fan-in versus fan-out conventions matter.

Context and aliases

Level: Intermediate. Stages: Training · Theory.

Tags: ML · DL · SLM · LLM.

Aliases: Kaiming initialization.

Related concepts

Concepts linking here

Original references

Open original record, export or annotate →