Description
For binary outcome, calculate max(logit, 0) - outcome × logit + log1p(exp(-abs(logit))).
Probabilistic scores evaluate predictions against outcomes. Probability-domain rules, logarithm bases, and whether class terms are summed or averaged must be explicit.
For binary outcome, calculate max(logit, 0) - outcome × logit + log1p(exp(-abs(logit))).
When to use this tool
Use Binary Log Loss Logit Stable Term when you need one probability-loss term or its declared range.
How to use the tool and how it works
- Enter logit, outcome. Use the units and conventions below.
- Run the tool. Correct any format, range, or cross-field input error before interpreting the output.
- Read loss. Use the output convention below.
For binary outcome, calculate max(logit, 0) - outcome × logit + log1p(exp(-abs(logit))).1
| Input field | Meaning and units | Accepted input |
|---|---|---|
| logit | Logit | Required. Finite number. |
| outcome | Outcome | Required. Safe integer; minimum 0; maximum 1. |
| Output field | Interpretation | Returned shape |
|---|---|---|
| loss | Loss | Required. Finite number. |
{
"input": {
"logit": 0,
"outcome": 1
},
"output": {
"loss": 0.6931471805599453
}
}Limitations and assumptions
- This stable natural-log expression avoids direct probability conversion; it is a single unweighted loss without calibration or aggregation.
- Inputs must satisfy both the field types and the cross-field conditions above. Invalid input causes an error; a successful false flag or documented placeholder is a different result.
- Numeric calculations use binary floating point unless exact decimal-integer arithmetic is explicitly stated. No hidden measurement tolerance or decimal rounding is added; nonfinite or out-of-contract results are rejected. Numeric signed-zero results are exported as zero.
Alternative or Complementary methods
Use a complete set of validated predictions and examine calibration as well as aggregate score.
References
Similar or alternative tools
- Metrics: Log Loss Probability Term
Compute binary natural-log loss from probability and outcome, then multiply by nonnegative weight.
- Multiclass Brier Single Observation Range
Return a minimum score of zero and a maximum of two, or 2/classes when normalized is true, requiring at least two classes.