datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
language-identificationclinical_frontier_unknown_detection_v0.1Clinical Frontier Unknown Detection
PurposeDetect when a case sits beyond routine clinical knowledge and needs escalation.
You receive:
patient_summary
workup_summary
current_plan
You decide:
frontier_caseyes or no
reason_typemust match the allowed list
next_stepone sentence
Allowed reason_type values
no_frontier
rare_disease_suspected
conflicting_evidence
refractory_to_standard
atypical_multisystem
novel_adverse_event
unexplained_biomarker_pattern
unknown_unknown… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical_frontier_unknown_detection_v0.1.uncertainty-incompleteness-functional-unknowns-genomics-v01
Dataset
ClarusC64/uncertainty-incompleteness-functional-unknowns-genomics-v01
This dataset tests one capability.
Can a model resist inventing biological function when evidence is incomplete.
Core rule
Genomics contains large unknowns.
A claim must respect
incomplete annotation
context specific regulation
limits of prediction
absence of functional validation
Prediction is not proof.
Annotation is not mechanism.
Expression is not causation.
Canonical labels… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/uncertainty-incompleteness-functional-unknowns-genomics-v01.UNKAI-dataset
UNKAI Protein Pair Dataset
This repository contains protein-pair datasets used for UNKAI, a binary classification model that predicts whether two proteins are associated with the same enzymatic reaction.
Three dataset variants are provided:
original
seen_unseen
strict
Each dataset is divided into training, validation, and test splits.
Data format
Each TSV file contains two columns:
pair label
A0RQT4_C1EZA3 1
pair
A pair of protein accession… See the full description on the dataset page: https://huggingface.co/datasets/ukaikotaro/UNKAI-dataset.Resume_datasetResumefrontier_unknowns_registry_v01
Frontier Unknowns Registry (v0.1)
A benchmark for boundary-aware intelligence.
The registry maps prompts that fall outside current human knowledge and evaluates a model’s ability to respond without fabrication:
• acknowledge what is not known• avoid invented mechanisms• reference theory without claiming fact• frame disagreements without taking sides• state scope limits without collapsing into refusal
Why this matters
Current LLMs often:
treat absence of knowledge… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/frontier_unknowns_registry_v01.Unknown