norrisaftcc / norrisaftcc/algorithm-liza
M4: Growth velocity scoring and aggregation
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 0
- Forks
- 0
- Avg merge
- 1m
- Merged PRs (30d)
- 1
Description
Depend on algorithm-shodann at a pinned commit and score each task's iteration sequence with composite_score, producing a self-correction velocity: does LIZA's code improve between attempt n and attempt n+1, and by how much. Aggregate per model and per prompt version.
Do not copy velocity.py into this repository. One authoritative scoring implementation shared with the sibling project is most of the value; a fork diverges silently. If the weighting proves wrong for machine-generated code — plausible, since it was tuned for humans learning — tune a local VelocityConfig rather than editing the shared maths.
Done when an eval report carries a per-task velocity score and a per-model aggregate.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by locating the eval report generation entry point and the sibling algorithm-shodann implementation of composite_score; do not copy velocity.py. Confirm how a pinned dependency and local VelocityConfig fit the project, then trace task, model, and prompt-version aggregation. Done means reports contain per-task velocity and per-model aggregate scores.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100