Rethinking the Initialization of Implicit Neural Representations (INRs)

A starting point

Rethinking the Initialization of Implicit Neural Representations (INRs) is one series within this broader research notebook.

This series will collect questions about initialization, notes from related reading, and records of experiments. The aim is to make the reasoning behind each step easy to revisit.

What belongs here

• Questions and hypotheses: observations worth investigating and assumptions that need testing.
• Reading notes: the motivation, methods, and limitations of related work.
• Experiment records: configurations, random seeds, observations, and reproducible code.

An ongoing notebook

This opening note introduces the series. Specific methods and conclusions will be developed in later posts, alongside their supporting experiments and references.