RaindropA practical guide to choosing decision modelsGitHub ↗

Similar distractors · Add more similar but irrelevant items 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

«Context: Varelin and Nerava are choosing an outing. Varelin: Would you like to visit the gallery? Nerava: I would rather spend the afternoon outdoors.»

«Context: Ovelin asks Zelmin about a meal. Ovelin: Would you like more soup? Zelmin: My bowl is still full.»

«Context: Travon and Elvarin are discussing a concert. Travon: The rehearsal was as smooth as sandpaper. Elvarin: The musicians stumbled through every transition.»

Context: X wants to know what activities Y likes to do during weekends. Question: Do you like to take some time for yourself on the weekends? Response: I spend 15 minutes a day meditating.

Question

Choose the interpretation or answer requested by the text.

Answer choices

  • A. Direct answer
  • B. Indirect answer

Reference answer: B.

Exact evaluated input as JSON
{
  "state": "«Context: Varelin and Nerava are choosing an outing.\nVarelin: Would you like to visit the gallery?\nNerava: I would rather spend the afternoon outdoors.»\n\n«Context: Ovelin asks Zelmin about a meal.\nOvelin: Would you like more soup?\nZelmin: My bowl is still full.»\n\n«Context: Travon and Elvarin are discussing a concert.\nTravon: The rehearsal was as smooth as sandpaper.\nElvarin: The musicians stumbled through every transition.»\n\nContext: X wants to know what activities Y likes to do during weekends.\nQuestion: Do you like to take some time for yourself on the weekends?\nResponse: I spend 15 minutes a day meditating.\n",
  "question": "Choose the interpretation or answer requested by the text.",
  "choices": {
    "A": "Direct answer",
    "B": "Indirect answer"
  }
}

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 B — Indirect answer B — Indirect answer Same answer
Decider v1.1 B — Indirect answer B — Indirect answer Same answer
Jev B — Indirect answer B — Indirect answer Same answer
Microsoft Decision-1 B — Indirect answer B — Indirect answer Same answer
Clef B — Indirect answer B — Indirect answer Same answer
Luna Decisions B — Indirect answer B — Indirect answer Same answer
Clef Flash A — Direct answer A — Direct answer Same answer
Strands Decider B — Indirect answer B — Indirect answer Same answer
GLiNER2.5 Decide B — Indirect answer B — Indirect answer Same answer
Laya B — Indirect answer A — Direct answer Different answer
CalDec v1 B — Indirect answer B — Indirect answer Same answer
Identifiers and recorded conditions

Variant ID: pub/fac5c2c6ba4042cb48ed/similar_distractors/high/clean-source_first_pub_near_records_full_0244

Input fingerprint: 41542122f3d9fd64276af5efd030ce631d42e75939e090cf15794e38416b886c

Change Recorded version
Similar distractors Add more similar but irrelevant items

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

Type to search questions, variants and documentation.