Skip to content

Explore smooth importance-weighted decay half-life #49

Description

@Atharva-Kanherkar

Summary

The current decay implementation uses a hard threshold: memories with importance >= 0.8 get a 3x half-life multiplier, while 0.79 and below get the base half-life.

Why revisit this

This is simple and testable for v1, but it creates a discontinuity at the boundary. A smooth curve could let importance continuously modulate half-life while preserving deterministic behavior.

Proposed follow-up

  • evaluate one or two candidate curves for mapping importance -> half-life multiplier
  • define the desired behavior at importance 0.0, 0.5, 0.8, and 1.0
  • replace the threshold-specific test coverage with shape-based coverage for the chosen curve
  • confirm whether semantic, episodic, and procedural memories should all share the same importance curve

Notes

This should be a follow-up to issue #42, not part of its current acceptance criteria.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions