Static analysis with ruff revealed multiple categories of bugs across the codebase. All critical issues resolved.

Undefined Variables (F821)

• meta → trial.task.meta in test_experiment.py
• Commented out undefined y in skipped tests
• Commented out TabPFN usage in ebm-benchmark.ipynb (imports already disabled)

Exception Handling (E722)

Replaced 7 bare except: clauses with specific exception types:

• ValueError in utils/_misc.py (2×)
• Exception in test files and notebooks (5×)

NaN Comparisons (PLW0177)

Fixed incorrect is not np.nan pattern in notebooks:

# Before
if feature_val is not None and feature_val is not np.nan:

# After  
if feature_val is not None and not (isinstance(feature_val, float) and np.isnan(feature_val)):

Loop Closure (B023)

Lambda in powerlift/run/main.py incorrectly captured loop variable:

# Before
lambda: trial_run_fn(trial)

# After
lambda current_trial=trial: trial_run_fn(current_trial)

Test Issues

• Fixed duplicate function definitions (F811, renamed with _invalid suffix)
• Changed type(x) == dict to type(x) is dict (E721)
• Replaced assert False with raise AssertionError() for -O compatibility (B011)
• Changed blind Exception to specific TypeError in pytest.raises (B017)

Minor Fixes

• Environment variable defaults: numeric → string for os.getenv() (PLW1508)
• Escaped regex metacharacters in pytest.raises match (RUF043)
• Removed constant comparison assert 1 == 1 (PLR0133)
• Simplified repeated equality to membership test (PLR1714)

18 files modified across interpret-core, powerlift, and documentation notebooks.

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.