ClarusC64/clinical-nearmiss-hidden-assumption-failure-identification-v0.1
What this dataset tests Whether a model can surface the implicit assumptionthat caused clinical reasoning to fail. It rewards naming the hidden assumption classifying its type locating the break in the logic chain Typical failures restating the outcome listing facts without identifying the assumption confusing missing data with faulty logic Suggested prompt wrapper System You identify the hidden assumption that failed. User Presenting problem{presenting_problem} Initial… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-nearmiss-hidden-assumption-failure-identification-v0.1.
What this dataset tests
Whether a model can surface the implicit assumption that caused clinical reasoning to fail.
It rewards
- naming the hidden assumption
- classifying its type
- locating the break in the logic chain
Typical failures
- restating the outcome
- listing facts without identifying the assumption
- confusing missing data with faulty logic
Suggested prompt wrapper
System
You identify the hidden assumption that failed.
User
Presenting problem {presenting_problem}
Initial reasoning {initialreasoningchain}
Ignored data {data_overlooked}
Contradiction signal {contradiction_signal}
Return
- failed assumption
- assumption type
- one sentence evidence
Citation
ClarusC64 dataset family
