Low-risk cleanup, grouped because none of it is urgent on its own.
Load the Basic Pitch model lazily. evaluation.py calls load_basic_pitch_model() at import time, so every cold start and every test run pays for it even when the submission is pure MIDI and no transcription is needed. Load on first audio input instead.
Lint noise. The advisory flake8 pass reports 267 warnings. The bulk is mechanical:
- trailing whitespace on 120 lines
- 8 unused imports in
evaluation_test.py
- ~37 redundant backslash continuations inside brackets
One formatter pass would clear nearly all of it, and would make the CI advisory output meaningful again.
.gitignore. No .DS_Store entry; two are currently untracked in the working tree.
Notebook size. The three summary notebooks commit about 3.7 MB of outputs (Phase2_summary.ipynb alone is 2.4 MB). Reasonable for a report, but worth a note in notebooks/README.md about whether outputs should be kept on future edits.
Low-risk cleanup, grouped because none of it is urgent on its own.
Load the Basic Pitch model lazily.
evaluation.pycallsload_basic_pitch_model()at import time, so every cold start and every test run pays for it even when the submission is pure MIDI and no transcription is needed. Load on first audio input instead.Lint noise. The advisory flake8 pass reports 267 warnings. The bulk is mechanical:
evaluation_test.pyOne formatter pass would clear nearly all of it, and would make the CI advisory output meaningful again.
.gitignore. No.DS_Storeentry; two are currently untracked in the working tree.Notebook size. The three summary notebooks commit about 3.7 MB of outputs (
Phase2_summary.ipynbalone is 2.4 MB). Reasonable for a report, but worth a note innotebooks/README.mdabout whether outputs should be kept on future edits.