7a92652289
Set up project policy files, baseline best-practice tracking, and a pre-commit performance gate so future train.py commits require measured RMSD improvement. Made-with: Cursor
36 lines
831 B
Markdown
36 lines
831 B
Markdown
# GUIDELINES
|
|
|
|
## Purpose
|
|
|
|
Make overfitting robust and measurable, targeting `mean_rmsd_100 <= 1.0`.
|
|
|
|
## Workflow
|
|
|
|
1. Modify code/config.
|
|
2. Run training/evaluation and write `reports/latest_eval.json`.
|
|
3. If improved, update `BEST_PRACTICE.json` in the same commit.
|
|
4. Append one line to `README.md` attempt log.
|
|
5. Commit.
|
|
|
|
## Required report format
|
|
|
|
`reports/latest_eval.json` must include:
|
|
|
|
- `mean_rmsd_100` (float, lower is better)
|
|
- `num_runs` (int, must be 100)
|
|
- `timestamp_utc`
|
|
- `command`
|
|
- `notes`
|
|
|
|
## Repro notes
|
|
|
|
- Keep seed explicit in commands.
|
|
- Keep sample path explicit.
|
|
- Prefer additive experiments (do not silently remove prior working options).
|
|
|
|
## Safety
|
|
|
|
- If pre-commit blocks due to no improvement on `train.py`, either:
|
|
- improve model and re-evaluate, or
|
|
- commit non-`train.py` changes separately.
|