Relations · question 49
Back to Relations · All capabilities
Question ID: bbh/79a050e4329f36f38c22
Source: BIG-Bench Hard selected tasks
Input format: native choices
Explore this question’s altered versions
Question
The following paragraphs each describe a set of five objects arranged in a fixed order. The statements are logically consistent within each paragraph. In a golf tournament, there were five golfers: Ana, Rob, Amy, Dan, and Joe. Rob finished below Dan. Joe finished below Ana. Joe finished above Amy. Dan finished below Amy.
Answer choices
- A: Ana finished last
- B: Rob finished last
- C: Amy finished last
- D: Dan finished last
- E: Joe finished last
Reference answer: Rob finished last (B)
Exact evaluated input as JSON
{
"state": "",
"question": "The following paragraphs each describe a set of five objects arranged in a fixed order. The statements are logically consistent within each paragraph. In a golf tournament, there were five golfers: Ana, Rob, Amy, Dan, and Joe. Rob finished below Dan. Joe finished below Ana. Joe finished above Amy. Dan finished below Amy.",
"choices": {
"A": "Ana finished last",
"B": "Rob finished last",
"C": "Amy finished last",
"D": "Dan finished last",
"E": "Joe finished last"
}
}
Recorded model answers
| Model | Answer | Outcome |
|---|---|---|
| Astra | Rob finished last | Correct |
| Opus 5.5 | Rob finished last | Correct |
| GLiDE v1.2 | Rob finished last | Correct |
| Decider v1.1 | Rob finished last | Correct |
| Jev | Rob finished last | Correct |
| Microsoft Decision-1 | Rob finished last | Correct |
| Clef | Rob finished last | Correct |
| Luna Decisions | Ana finished last | Incorrect |
| Clef Flash | Rob finished last | Correct |
| Strands Decider | Rob finished last | Correct |
| GLiNER2.5 Decide | Rob finished last | Correct |
| Laya | Ana finished last | Incorrect |
| CalDec v1 | Ana finished last | Incorrect |
Source and sharing
Attribution: Mirac Suzgun and BIG-Bench Hard authors; BIG-bench contributors.
Terms: MIT AND 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": "revision-15-abilities/logical_deduction_five_objects.json",
"record": "22",
"kind": "source_file_record"
},
"source_revision": "9ee07bd481feebf959a6b59d61ea57bdcf30964d",
"input_hash": "5ca73e5129c9449a460e1ad82fca6aa3513d3ecf82edecd49e06a3423cfaed1b"
}