RaindropA practical guide to choosing decision modelsGitHub ↗

Learning and induction · question 7

Back to Learning and induction · All capabilities

Question ID: list_functions/3e2239ccffbdb2c9f6ba
Source: List Functions
Input format: adapted choices

Explore this question’s altered versions

Context

[{"input": "[3, 5, 4]", "output": "[5, 7, 6]"}, {"input": "[7, 5, 3, 6, 1, 4, 0]", "output": "[9, 7, 5, 8, 3, 6, 2]"}, {"input": "[1, 5, 4, 7, 0]", "output": "[3, 7, 6, 9, 2]"}, {"input": "[5, 3, 0, 1]", "output": "[7, 5, 2, 3]"}, {"input": "[6, 5, 7]", "output": "[8, 7, 9]"}, {"input": "[7, 6, 0, 3]", "output": "[9, 8, 2, 5]"}, {"input": "[1, 0, 6, 3, 2, 5]", "output": "[3, 2, 8, 5, 4, 7]"}, {"input": "[3, 7, 6, 1]", "output": "[5, 9, 8, 3]"}]

Question

Apply the same inferred rule to this new input: [2]

Answer choices

  • A: [4]
  • B: [1]
  • C: [3]
  • D: [2]

Reference answer: [4] (A)

Exact evaluated input as JSON
{
  "state": "[{\"input\": \"[3, 5, 4]\", \"output\": \"[5, 7, 6]\"}, {\"input\": \"[7, 5, 3, 6, 1, 4, 0]\", \"output\": \"[9, 7, 5, 8, 3, 6, 2]\"}, {\"input\": \"[1, 5, 4, 7, 0]\", \"output\": \"[3, 7, 6, 9, 2]\"}, {\"input\": \"[5, 3, 0, 1]\", \"output\": \"[7, 5, 2, 3]\"}, {\"input\": \"[6, 5, 7]\", \"output\": \"[8, 7, 9]\"}, {\"input\": \"[7, 6, 0, 3]\", \"output\": \"[9, 8, 2, 5]\"}, {\"input\": \"[1, 0, 6, 3, 2, 5]\", \"output\": \"[3, 2, 8, 5, 4, 7]\"}, {\"input\": \"[3, 7, 6, 1]\", \"output\": \"[5, 9, 8, 3]\"}]",
  "question": "Apply the same inferred rule to this new input: [2]",
  "choices": {
    "A": "[4]",
    "B": "[1]",
    "C": "[3]",
    "D": "[2]"
  }
}

Recorded model answers

Model Answer Outcome
Astra [4] Correct
Opus 5.5 [4] Correct
GLiDE v1.2 [4] Correct
Decider v1.1 [4] Correct
Jev [4] Correct
Microsoft Decision-1 [4] Correct
Clef [4] Correct
Luna Decisions [4] Correct
Clef Flash [4] Correct
Strands Decider [3] Incorrect
GLiNER2.5 Decide [2] Incorrect
Laya [3] Incorrect
CalDec v1 [1] Incorrect

Source and sharing

Licensed upstream task: c071. All eight supplied examples and the evaluated test input/output match the original BIG-bench data.

Attribution: Joshua S. Rule, Steven T. Piantadosi, Joshua B. Tenenbaum; BIG-bench contributors; Linlu Qiu et al. (selection).
Terms: Apache-2.0. Source sharing decision.
The capability assignment and choice-format normalization are part of this experiment. Source data retain their own terms.

Exact source locator and input fingerprint
{
  "source_locator": {
    "path": "data/list_function.jsonl",
    "record": "70/4",
    "kind": "source_file_record"
  },
  "source_revision": "0eae21f3d3014b92be34c80b23c9120a7b15a51f",
  "input_hash": "73172972d5d8fab9a166a8a8632f0467bacce64bc521e495913d9cb7eec96813"
}

Type to search questions, variants and documentation.