Skip to content

Commit

Permalink
Interpretable Elo scores (#116)
Browse files Browse the repository at this point in the history
* Print more intuitive Elo values (consistent with the more common Elo scale)
* Changed wording to be consistent in CI logging.
  • Loading branch information
Eugenio-Bruno authored Dec 19, 2020
1 parent 9e4ef67 commit a41079b
Showing 1 changed file with 6 additions and 6 deletions.
12 changes: 6 additions & 6 deletions tune/cli.py
Original file line number Diff line number Diff line change
Expand Up @@ -371,15 +371,15 @@ def local( # noqa: C901
)
root_logger.info(f"Current optimum:\n{best_point_dict}")
root_logger.info(
f"Estimated value: {np.around(best_value, 4)} +- "
f"{np.around(best_std, 4).item()}"
f"Estimated Elo: {np.around(-best_value * 100, 4)} +- "
f"{np.around(best_std * 100, 4).item()}"
)
confidence_val = settings.get("confidence", confidence)
confidence_mult = erfinv(confidence_val) * np.sqrt(2)
root_logger.info(
f"{confidence_val * 100}% confidence interval of the value: "
f"({np.around(best_value - confidence_mult * best_std, 4).item()}, "
f"{np.around(best_value + confidence_mult * best_std, 4).item()})"
f"{confidence_val * 100}% confidence interval of the Elo value: "
f"({np.around(-best_value * 100 - confidence_mult * best_std * 100, 4).item()}, "
f"{np.around(-best_value * 100 + confidence_mult * best_std * 100, 4).item()})"
)
confidence_out = confidence_intervals(
optimizer=opt,
Expand Down Expand Up @@ -457,7 +457,7 @@ def local( # noqa: C901
root_logger.info(f"Experiment finished ({difference}s elapsed).")

score, error_variance = parse_experiment_result(out_exp, **settings)
root_logger.info("Got score: {} +- {}".format(score, np.sqrt(error_variance)))
root_logger.info("Got Elo: {} +- {}".format(-score * 100, np.sqrt(error_variance) * 100))
root_logger.info("Updating model")
while True:
try:
Expand Down

0 comments on commit a41079b

Please sign in to comment.