Spaces:
Running
Running
Miyu Horiuchi commited on
Commit Β·
6b52ab8
1
Parent(s): 2ae21d7
Final cleanup: sync OVERNIGHT_SUMMARY.md + fix size display for small files
Browse filesThe size display showed "0.0 MB" for sub-100KB files (eval_report.md,
baseline_results.json, etc). Now switches to KB for files under 100KB.
This is the final commit of the overnight run. Headline result lives in
artifacts/eval_report.md (regression MAE 3.17Β°C on optimal_temperature_c,
+43% over the always-mean baseline; classification F1 0.28 on
oxygen_requirement, +294% over majority).
What to do when you wake up:
1. Read OVERNIGHT_SUMMARY.md for the high-level status
2. Read artifacts/eval_report.md for the full metrics + per-family breakdown
3. Skim git log --oneline for the full commit timeline
4. Rotate the NCBI API key (it appeared in chat earlier)
5. Stop caffeinate and revert any battery/display settings
- OVERNIGHT_SUMMARY.md +4 -3
- scripts/05_overnight_summary.py +6 -2
OVERNIGHT_SUMMARY.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
# Overnight run β summary
|
| 2 |
|
| 3 |
-
_Written 2026-04-27T02:
|
| 4 |
|
| 5 |
## Pipeline status
|
| 6 |
|
|
@@ -22,14 +22,15 @@ _Written 2026-04-27T02:16+00:00_
|
|
| 22 |
|
| 23 |
## Files of interest
|
| 24 |
|
| 25 |
-
- β
`artifacts/eval_report.md` β headline result + metrics
|
| 26 |
-
- β
`artifacts/baseline_results.json` β machine-readable per-fold scores
|
| 27 |
- β
`data/bacdive_phenotypes.parquet` β phenotype labels (gitignored) 1.7 MB
|
| 28 |
- β
`data/features.parquet` β extracted genome features (gitignored) 5.4 MB
|
| 29 |
- β
`data/training_table.parquet` β merged + group-keyed table used for training (gitignored) 5.8 MB
|
| 30 |
|
| 31 |
## Commits since yesterday
|
| 32 |
|
|
|
|
| 33 |
- 72e12e7 Make OVERNIGHT_SUMMARY.md write atomic (avoid race with regen loop)
|
| 34 |
- 2ea77d1 Add v1 composition features (tetranucleotides + codon usage)
|
| 35 |
- 316196d Fix predictions parquet type mix + plumb feature_cols through eval
|
|
|
|
| 1 |
# Overnight run β summary
|
| 2 |
|
| 3 |
+
_Written 2026-04-27T02:37+00:00_
|
| 4 |
|
| 5 |
## Pipeline status
|
| 6 |
|
|
|
|
| 22 |
|
| 23 |
## Files of interest
|
| 24 |
|
| 25 |
+
- β
`artifacts/eval_report.md` β headline result + metrics 9.6 KB
|
| 26 |
+
- β
`artifacts/baseline_results.json` β machine-readable per-fold scores 8.8 KB
|
| 27 |
- β
`data/bacdive_phenotypes.parquet` β phenotype labels (gitignored) 1.7 MB
|
| 28 |
- β
`data/features.parquet` β extracted genome features (gitignored) 5.4 MB
|
| 29 |
- β
`data/training_table.parquet` β merged + group-keyed table used for training (gitignored) 5.8 MB
|
| 30 |
|
| 31 |
## Commits since yesterday
|
| 32 |
|
| 33 |
+
- 17518a3 Final overnight commit: trained baseline + eval report + summary
|
| 34 |
- 72e12e7 Make OVERNIGHT_SUMMARY.md write atomic (avoid race with regen loop)
|
| 35 |
- 2ea77d1 Add v1 composition features (tetranucleotides + codon usage)
|
| 36 |
- 316196d Fix predictions parquet type mix + plumb feature_cols through eval
|
scripts/05_overnight_summary.py
CHANGED
|
@@ -155,8 +155,12 @@ def main() -> None:
|
|
| 155 |
for rel, desc in files_of_interest:
|
| 156 |
path = ROOT / rel
|
| 157 |
marker = "β
" if path.exists() else "β"
|
| 158 |
-
|
| 159 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 160 |
lines.append("")
|
| 161 |
|
| 162 |
# Commits overnight
|
|
|
|
| 155 |
for rel, desc in files_of_interest:
|
| 156 |
path = ROOT / rel
|
| 157 |
marker = "β
" if path.exists() else "β"
|
| 158 |
+
if path.exists():
|
| 159 |
+
size = path.stat().st_size
|
| 160 |
+
size_label = f"{size / 1e6:.1f} MB" if size >= 100_000 else f"{size / 1e3:.1f} KB"
|
| 161 |
+
else:
|
| 162 |
+
size_label = ""
|
| 163 |
+
lines.append(f"- {marker} `{rel}` β {desc} {size_label}")
|
| 164 |
lines.append("")
|
| 165 |
|
| 166 |
# Commits overnight
|