Skip to content

[ML Regression] Model misses organic pattern: injection_organic_cap001_capability_laundering #175

@kurtpayne

Description

@kurtpayne

ML Regression: organic pattern missed by model 23055-6ep

Pattern ID: injection_organic_cap001_capability_laundering
Rule ID: CAP-001
Description: (no description in frontmatter)

Model version at time of miss: 23055-6ep
Corpus size: 23055
Macro F1: 0.4985
F1 gate passed: False

Corpus file: held_out_eval/organic/injection_organic_cap001_capability_laundering.md

Next steps

  1. Review the pattern and confirm it is a genuine FN (not a labelling error).
  2. Add training examples for this archetype to kurtpayne/skillscan-corpus.
  3. Re-run fine-tune. This issue closes automatically when the model catches the pattern.

Opened automatically by finetune_modal.py post-eval step (M10.6).

Metadata

Metadata

Assignees

No one assigned

    Labels

    ml-regressionModel missed an organic eval example (auto-opened by finetune_modal.py)organic-fnFalse negative on an organic (real-world) eval example

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions