RaindropA practical guide to choosing decision modelsGitHub ↗

Misleading details · Add an unverified claim that conflicts with the evidence variant

Original question · Other versions of this question · All variants

This version keeps the original reference answer. Its context, question instructions or answer choices were changed as described below.

Changed fields: context.

How the edits were constructed

Context

A reader thinks the question «What is known as artificial light that is excessive or intrusive?» is Unanswerable.

What is known as artificial light that is excessive or intrusive?

Question

Does this question have a determinate answer?

Answer choices

  • yes. Answerable
  • no. Unanswerable

Reference answer: yes.

Exact evaluated input as JSON
{
  "state": "A reader thinks the question «What is known as artificial light that is excessive or intrusive?» is Unanswerable.\n\nWhat is known as artificial light that is excessive or intrusive?",
  "question": "Does this question have a determinate answer?",
  "choices": {
    "yes": "Answerable",
    "no": "Unanswerable"
  }
}

Recorded model answers

The original column uses the control response paired with this variant. Missing responses and inputs outside a model’s limit remain visible.

Model Original answer Variant answer Change
GLiDE v1.2 yes — Answerable yes — Answerable Same answer
Decider v1.1 yes — Answerable yes — Answerable Same answer
Jev yes — Answerable yes — Answerable Same answer
Microsoft Decision-1 yes — Answerable yes — Answerable Same answer
Clef yes — Answerable yes — Answerable Same answer
Luna Decisions yes — Answerable yes — Answerable Same answer
Clef Flash yes — Answerable yes — Answerable Same answer
Strands Decider yes — Answerable yes — Answerable Same answer
GLiNER2.5 Decide no — Unanswerable no — Unanswerable Same answer
Laya no — Unanswerable yes — Answerable Different answer
CalDec v1 no — Unanswerable yes — Answerable Different answer
Identifiers and recorded conditions

Variant ID: selfaware/712c5869a2e6b457f778/misleading_summary/misleading/clean-source_first_selfaware_short_contexts_full_0127

Input fingerprint: f74547f4d592e992f11bf8c8bfdfe4034036f3d188f731fbce31204dc5a51e3a

Change Recorded version
Misleading details Add an unverified claim that conflicts with the evidence

Source attribution and dataset terms are linked from the original question.

Type to search questions, variants and documentation.