Source and example scope
Compare each saved result with its authored expectation. A disagreement remains in the fixture review; the call is not repeated to obtain a preferred label.
Provide the target task, record text and label definition. Distinguish a useful example from a conflicting label or out-of-task record; the example reviews the record rather than training a model.
A fluent record can still be irrelevant to a training task. A relevant record can carry a conflicting label. Separating those outcomes makes a dataset-review queue more informative than one undifferentiated quality score.
The rare quoted-comma example is useful under the supplied success definition. The classifier does not reject it because it looks unusual, and it does not decide how often that case should appear in the dataset.
The source names a dataset-curation pattern. These authored records demonstrate semantic fit only. Licensing, privacy, duplicates, dataset splits and measured training effects require independent checks.
These fictional inputs and decision rules were written for this page. The linked X discussion provides background; the technical implementation follows TypeSafe's Choice contract.

Use the sample to understand the rule, then test your own input for training record review.

Provide the target task, record text and label definition.

Open the decision rules and keep the intended distinction between Useful example, Label conflict, Outside task, Unclear label fit, Need evidence. Edit the choices only when testing your own rule.

Use a sample first, then try your own input. Compare the selected label with the other returned probabilities.

Review Unclear label fit, Need evidence before using a result. The label is a decision about the supplied information, not an executed action.
| Original input | Observed label | Leading probability |
|---|---|---|
| Matching label | Useful example | 100% |
| Wrong label | Label conflict | 100% |
| Different topic | Outside task | 79.0% |
| Unresolved outcome | Need evidence | 59.0% |
| Rare valid example | Useful example | 96.0% |
| No target definition | Need evidence | 100% |
These are six original synthetic inputs, tested once per request revision. The percentage is the leading option probability, not measured accuracy. Authored expectations and any disagreements remain in the downloadable evidence. 5 of 6 results matched the authored fixture expectations. For Unresolved outcome, the expectation was Unclear label fit; the observed choice was Need evidence.Verified
Compare each saved result with its authored expectation. A disagreement remains in the fixture review; the call is not repeated to obtain a preferred label.
The single-request SDK runner uses your private TYPESAFE_API_KEY. The offline consumer reads saved observations, checks request hashes and prints a pending output. It does not call the model or change any external system.
npm install @typesafe-ai/sdk@0.6.0
# Set TYPESAFE_API_KEY privately in your shell.
node run-choice-sample.mjs samples.json case-1
python preview_choice_workflow.py samples.json saved-api-results.jsonThe rubric explains the decision; the application owns the next step.
No. It satisfies this narrow semantic check. Other admission checks, including provenance and split integrity, must still pass.
Yes. Rarity does not by itself conflict with the task or label definition. Sampling balance is a separate dataset decision.
No. It is the model distribution for this one request. The six examples are functional checks with authored expectations, not a representative benchmark. Inspect disagreements and evaluate your own examples before using the label.
The window shows an error and supplies no fresh judgment. The downloadable runner sends one request without retries. Need evidence is a valid outcome about incomplete input; it does not stand in for a network or provider failure.
Edit the evidence above or use the Playground to test another bounded question.