Use scientific notation for best error in logger output

[Copilot]

Logger output displays best error=0.0000 for very small error values (e.g., 1e-07), losing precision entirely due to fixed-point formatting with 4 decimal places.

Changes

• Changed format specifier from {:.4f} to {:.4e} in flaml/automl/automl.py:2973 for both search_state.best_loss and self._state.best_loss

Impact

Before:

at 5.2s,  estimator lgbm's best error=0.0000,  best estimator lgbm's best error=0.0000

After:

at 5.2s,  estimator lgbm's best error=1.2300e-06,  best estimator lgbm's best error=1.2300e-06

Scientific notation preserves precision across all error magnitudes while maintaining 4 significant figures.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Represent "best error" in runtime logger instance using scientific notation?</issue_title>
<issue_description>My logger output is full of "best error=0.0000" and I suspect there are digits beyond this which scientific notation could show properly.</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes Represent "best error" in runtime logger instance using scientific notation? #1272

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.