Jev experiment model attribution

Check which model an experiment result actually describes

Give the result claim and the systems included in the method. Compare attribution to a tested system with an unsupported transfer or an explicit architecture analogy.

Was the model named in the claim tested?Live API
Try an example

Measured claim remains with the tested system.

Source: API-collected workflow inspiration; original demonstration below

Source and example scope

Original attribution-review adaptation of a source that explicitly distinguishes its open-model experiments from Jev API testing. Its reported results remain author claims. The six editable inputs and outcome rules are original synthetic demonstrations.

Compare the result claim with the tested systems.

Decision rules and labels
Attribution matchesAttribution expandedArchitecture analogyNeed method
A real response from the Jev API

Edit any input to explore a different decision. No request is sent until you run it.

Your next decision

Not run yet

Choose an example or write your own input, then run it to see the ranked choices.

Results are not stored by this site. Your submitted text is processed by the API provider.

The decision

An API-compatible implementation is still a different test subject.

The transferred-result preset gives Jev a speed result from an experiment that tested only the fictional CedarSmall model. The claim changes the owner of the measurement, even though the method explicitly names its actual subject.

An architecture analogy can be useful without transferring a result. The analogy preset compares a decision-scoring idea while keeping every measurement attributed to CedarSmall. That boundary differs from saying both systems achieved the same outcome.

This demonstration checks attribution in supplied text. It does not verify the experiment, measure speed or authenticate a report. The fictional examples are original tests of this rule; their invented result claims are not Jev benchmark findings.

A tested system name does not prove that a particular metric belongs to that system when several models were evaluated. That needs a separate result-to-model mapping.

Original attribution-review adaptation of a source that explicitly distinguishes its open-model experiments from Jev API testing. Its reported results remain author claims. The six editable inputs and outcome rules are original synthetic demonstrations.

Illustration of matching a question to supplied passages and identifying missing evidence
Illustration of the decision flow.
How it works

Try the example in four steps

Use the sample to understand the rule, then test your own input for experiment result attribution.

  1. Illustration of adding the text or evidence for a decision

    Add the information to check

    Give the result claim and the systems included in the method.

  2. Illustration of defining a question and its allowed choices

    Confirm the question and labels

    Open the decision rules and keep the intended distinction between Attribution matches, Attribution expanded, Architecture analogy, Need method. Edit the choices only when testing your own rule.

  3. Illustration of sending one input and comparing the returned choices

    Run the decision

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

  4. Illustration of reviewing a result and an uncertain item before using it

    Review the next step

    Review Need method before using a result. The label is a decision about the supplied information, not an executed action.

Our API test

Recorded API observations

Recorded API observations
Original inputObserved labelLeading probability
Correct test subjectAttribution matches99.0%
Result transferredAttribution expanded100%
Explicit analogyAttribution matches49.0%
Compatible is not identicalAttribution expanded97.0%
Ambiguous model nameNeed method98.0%
Missing methodNeed method94.0%

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 Explicit analogy, the expectation was Architecture analogy; the observed choice was Attribution matches.Verified

Understand the results

Source and example scope

Replay each saved synthetic request and compare the observed label with its authored expectation. Differences remain visible; these six inputs are functional examples, not a production benchmark.

Using the result in your application

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.

Runnable offline example

Run one sample, then inspect the local packet

View the code example
Single-request replay
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.json
Scope of this demo

Where this decision fits

Useful scope

  • Checking whether a result claim names a tested system
  • Separating mechanism analogies from measurement claims
  • Finding benchmark summaries that omit the tested identity

Outside this demonstration

  • Verifying a benchmark or calculating its results
  • Assuming compatible APIs are identical models
  • Inferring test methods that were not supplied
Example FAQ

Questions about experiment model attribution.

What the supplied evidence can establish, and what it leaves open.

Can a local model explain the same decision concept?

Yes, as an explicitly scoped analogy. Its measurements still belong to that local model and do not become Jev API results.

Does Attribution matches verify the claimed result?

No. It establishes only that the claim refers to a system named in the supplied method. The experiment and its measurements still require their own evidence.

Is the highest probability an accuracy score?

No. It is the option distribution for this request. The saved examples compare model outputs with authored expectations and preserve any disagreement.

What happens if the API request fails?

The window shows an error and supplies no fresh judgment. Missing-evidence labels describe the input; they are not substitutes for network or provider errors.

Try your own experiment model attribution input

Edit the evidence in the window above, or open the Playground to create a different decision rule.