Skip to content

Commit 20bfc96

Browse files
Nick  VaccarelloNick  Vaccarello
authored andcommitted
formatting
1 parent 488896c commit 20bfc96

1 file changed

Lines changed: 3 additions & 0 deletions

File tree

medical_diagnosis_model/NEXT_STEPS.md

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -105,8 +105,10 @@ Acceptance:
105105
## Foundation upgrades (week 2–3)
106106

107107
- Splits & imbalance
108+
108109
- Patient‑level and time‑based train/val/test split (no leakage); stratify; add class weights or weighted sampling.
109110
- Acceptance criteria:
111+
110112
- `backend/data/splitter.py` implements patient/time‑based splits; no leakage tests pass.
111113
- CLI: `python -m backend.tools.split --input data/clean/ --out data/splits/` writes CSV lists and a class distribution report.
112114

@@ -143,6 +145,7 @@ Acceptance:
143145
- Update README/NEXT_STEPS with commands:
144146
- `PYTHONPATH=. python3 -m medical_diagnosis_model.backend.tools.split --strategy stratified`
145147
- `PYTHONPATH=. python3 medical_diagnosis_model/tools/train_pipeline.py --jsonl data/splits/v02/train.jsonl --epochs 5000`
148+
146149
- Training
147150

148151
- Switch to Adam; add L2 weight decay; optional dropout in hidden layer; expose via config.

0 commit comments

Comments
 (0)