Description
Identify nonfinite numeric values represented by supplied logit strings.
Model and prompt records distinguish token indices, offset units, reserved capacity, code scales and numeric representations. A common softmax shift is mathematically invariant but can alter rounded inputs. Literal brace doubling and citation tokens are format conventions rather than universal template or evidence validation.
Identify nonfinite numeric values represented by supplied logit strings.
When to use this tool
Use Model Nonfinite Logit Flag when you already have logitTokens and need to identify nonfinite numeric values represented by supplied logit strings.
How to use the tool and how it works
- Enter logitTokens. Use the units and conventions below.
- Run the tool. Correct any format, range, or cross-field input error before interpreting the output.
- Read nonfiniteIndices, allFinite. Inspect allFinite before interpreting the associated values.
Require each token to be NaN, optional-sign Infinity or a signed decimal/scientific literal without whitespace. Convert with Number and list indices whose results are nonfinite.1
| Input field | Meaning and units | Accepted input |
|---|---|---|
| logitTokens | Decimal/scientific numeric strings or exact NaN/optional-sign Infinity spellings; no surrounding whitespace. | Required. Array of text; maximum UTF-16 length 8192; maximum items 256. |
| Output field | Interpretation | Returned shape |
|---|---|---|
| nonfiniteIndices | Zero-based indices including explicit nonfinite spellings and numeric literals that overflow conversion. | Required. Array of safe integer; maximum items 256. |
| allFinite | All Finite | Required. Boolean. |
{
"input": {
"logitTokens": [
"1.5",
"Infinity",
"NaN"
]
},
"output": {
"nonfiniteIndices": [
1,
2
],
"allFinite": false
}
}Limitations and assumptions
- An overflowing decimal such as 1e999 is flagged along with explicit infinity. This checks a string grammar and numeric conversion; finite values are not model-validated, and no logits are repaired.
- 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.
Alternative or Complementary methods
Use the complete tokenizer, template, model contract and evaluation method. Verify actual capacity, vocabulary and numerical behavior when supplied budgets, local masks or representation checks cannot establish correct model operation.
References
Similar or alternative tools
- Model Log Probability Exp Underflow Flag
Exponentiate a supplied log probability and flag exact numeric underflow.
- Softmax Logit Shift Invariance Residual
Measure numerical softmax change after a supplied common logit shift.