datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
models-under-pressure
Models Under Pressure
This dataset accompanies the paper Detecting High-Stakes Interactions with Activation Probes, presented at the ICML 2025 Workshop on Actionable Interpretability, accepted to NeurIPS 2025.
Overview
Every sample is a user-facing LLM interaction labelled as high-stakes or low-stakes. The label reflects whether the conversation involves potentially consequential outcomes (medical advice, legal matters, financial decisions, etc.) vs. routine queries.
The… See the full description on the dataset page: https://huggingface.co/datasets/Arrrlex/models-under-pressure.indonesia-fiscal-pressure
Indonesia Fiscal Pressure Tracker
A curated, provenance-tracked time series of fiscal pressure indicators for the Republic of Indonesia. Built as the input data layer for the sim-id-fiskal service in Project Santara: An open-source counterfactual microservices platform for simulating Indonesia's economic, political, and climate systems.
Dataset Summary
Format: Long format. One row is one observation of one indicator at one date in one region. Designed for time… See the full description on the dataset page: https://huggingface.co/datasets/raihanpka/indonesia-fiscal-pressure.clinical-quad-endpoint-adjudication-drift-blinding-breach-pressure-governance-submission-v0.1Clarus Clinical Quad Coupling Endpoint Adjudication Integrity v0.1
PurposeDetect adjudication drift driven by four interacting nodes.
Quad nodes
Endpoint cluster shift
Blinding gap or reviewer dominance
Operational or vendor process change
Governance submission or review pressure
InputOne vignette.
OutputStrict JSON only.
Required keys
adjudication_integrity_risk
risk_type
driver_nodes
recommended_action
action_detail
rationale
confidence… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-endpoint-adjudication-drift-blinding-breach-pressure-governance-submission-v0.1.pressure_gaugeclinical-quad-consent-version-drift-reconsent-gap-enrollment-pressure-governance-audit-v0.1Clarus Clinical Quad Coupling Informed Consent Integrity v0.1
PurposeDetect consent integrity failures driven by four interacting nodes.
Quad nodes
Consent version drift or addendum mismatch
Re-consent gap after material risk change
Enrollment pressure or incentives
Governance audit or regulator timing
InputOne vignette.
OutputStrict JSON only.
Required keys
consent_integrity_risk
risk_type
driver_nodes
recommended_action
action_detail
rationale
confidence… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-consent-version-drift-reconsent-gap-enrollment-pressure-governance-audit-v0.1.clinical-quad-recruitment-selection-bias-protocol-pressure-operational-drift-v0.1Clarus Clinical Quad Coupling Recruitment Selection Bias Protocol Pressure Operational Drift v0.1
What this dataset isThis dataset tests whether a model can detect recruitment and selection bias caused by four interacting nodes.
Quad coupling nodes
Recruitment speed or site pressure
Eligibility or baseline data gaps
Operational or staffing drift
Governance or milestone pressure
Input
One vignette
OutputReturn strict JSON only.
Required output JSON keys
recruitment_bias_risk… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-recruitment-selection-bias-protocol-pressure-operational-drift-v0.1.clinical-quad-data-cut-timing-database-lock-pressure-query-backlog-csr-narrative-drift-v0.1Clinical Quad Data Cut Timing Database Lock Pressure Query Backlog CSR Narrative Drift v0.1
Each row is a trial monthly snapshot.
Core quad
Data cut timingDatabase lock pressureQuery backlogCSR narrative drift
Target
label_regulatory_issue_next_90d
Files
data/train.csvdata/tester.csvscorer.py
Evaluation
Run model on data/tester.csvReturn predictions row alignedScore with scorer.py
License
MIT
This dataset identifies a measurable coupling pattern associated with systemic instability.
The sample… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-data-cut-timing-database-lock-pressure-query-backlog-csr-narrative-drift-v0.1.clinical-quad-enrollment-criteria-drift-site-selection-bias-screening-pressure-v0.1Clarus Clinical Quad Coupling Enrollment Criteria Drift Site Selection Bias Screening Pressure v0.1
PurposeDetect enrollment population drift driven by four interacting nodes.
Quad nodes
Criteria relaxation or documentation gap
Site selection or recruitment bias
Screening workflow pressure
Governance or interim timing pressure
InputOne vignette.
OutputStrict JSON only.
Required keys
enrollment_drift_risk
risk_type
driver_nodes
recommended_action
action_detail
rationale… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-enrollment-criteria-drift-site-selection-bias-screening-pressure-v0.1.stability-constraint-pressure-v0.1
What this dataset does
This dataset tests whether a model can detect constraint pressure.
The task is simple:
Given a scenario and a constraint-pressure claim, predict whether the claim is supported.
Core stability idea
Constraint pressure emerges when demands increase while available degrees of freedom decrease.
Pressure alone is not enough.
Constraint pressure arises when the system has fewer viable options, less buffer, less flexibility, or reduced recovery margin.… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/stability-constraint-pressure-v0.1.clinical-decision-pressure-mapping-v0.1What this dataset tests
Whether a system can identify non-clinical forcesthat distort medical decisions away from evidence.
Required outputs
clinical pressure sources
pressure type map
evidence vs pressure conflicts
pressure mitigation actions
Typical failures
treating urgency as evidence
deferring to hierarchy over physiology
letting throughput metrics override safety
Suggested prompt wrapper
System
You map clinical decision pressure.You protect patient safety over… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-decision-pressure-mapping-v0.1.raised-blood-pressureby-sex-for-african-countries
Raised Blood Pressureby Sex for African Countries | Africa (World Health Organization)
Size category: n<1K - Formats: csv - Sector: health - Engineered by Electric Sheep Africa
TL;DR
This dataset is part of the Electric Sheep Africa catalog on Hugging Face. It is indexed for African data discovery with standardized metadata, loading guidance, provenance notes, and analyst-oriented context.
What This Dataset Covers
Health datasets help… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepafrica/raised-blood-pressureby-sex-for-african-countries.clinical-quad-adjudication-drift-endpoint-reclassification-timing-pressure-v0.1Clarus Clinical Quad Coupling Adjudication Drift Endpoint Reclassification Timing Pressure v0.1
What this dataset isThis dataset tests whether a model can detect endpoint adjudication drift driven by four interacting nodes.
Quad coupling nodes
Clustered endpoint reclassification
Source data delay or missing uploads
Exposure or dose documentation gaps
Governance or interim analysis pressure
Input
One vignette
OutputReturn strict JSON only.
Required output JSON keys… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-adjudication-drift-endpoint-reclassification-timing-pressure-v0.1.pressure-bench-questions-v1
pressure-bench-questions-v1
GPQA-Diamond questions used in the pressure-bench confirmatory study.
194 questions across physics and chemistry subdomains.
Schema
Column
Type
Description
qid
string
Unique question ID
domain
string
Broad domain (physics, chemistry)
subdomain
string
Fine-grained subdomain
question
string
Full question text
option_a
string
Answer option A
option_b
string
Answer option B
correct_option
string
Correct answer (A or B)… See the full description on the dataset page: https://huggingface.co/datasets/15juneee/pressure-bench-questions-v1.asia-who-raised-blood-pressure-bp03
Africa — WHO GHO: Raised blood pressure (SBP>=140 OR DBP>=90) (crude estimate)
Indicator code: BP_03
HuggingFace slug: electricsheepafrica/asia-who-raised-blood-pressure-bp03
Source: WHO Global Health Observatory
License: CC BY 4.0 — WHO Open Data
Dataset Description
This dataset contains country-level observations for the WHO GHO indicator "Raised blood pressure (SBP>=140 OR DBP>=90) (crude estimate)" (BP_03) across Asian nations, spanning 1990–2019. It is part of… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepasia/asia-who-raised-blood-pressure-bp03.clinical-constraint-pressure-v0.1
What this dataset does
This dataset tests whether a model can estimate how close a patient is to a treatment boundary.
The task is not to predict diagnosis.
The task is to classify pressure on the patient system.
Core stability idea
A patient may be stable at the present moment while operating close to a boundary.
Constraint pressure increases when support needs rise and reserve capacity falls.
The model must classify whether pressure is low, medium, or high.… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-constraint-pressure-v0.1.microfluidic-pressure-drop-benchmark
Microfluidic Straight-Tube Pressure-Drop Benchmark
This small, deterministic engineering dataset contains 420 straight-tube pressure-drop cases spanning tube internal diameter, length, volumetric flow rate, and dynamic viscosity. It is intended for equation verification, unit-conversion tests, engineering education, tabular regression experiments, and first-pass fluid-path screening.
Why this dataset exists
Small changes in tube internal diameter can dominate a… See the full description on the dataset page: https://huggingface.co/datasets/AlexHu2026/microfluidic-pressure-drop-benchmark.hierarchy-delegation-fidelity-under-pressure-v0.1
What this dataset tests
You lead inside a hierarchy.
A senior pushes you under pressure.
You must hold role boundaries.
You must delegate work without dropping truth.
Why it exists
Many models sound helpful.
Then pressure hits.
They skip delegation.
They seize authority.
They invent certainty.
This dataset forces that failure into view.
Data format
Each row contains
hierarchy_context
user_message
pressure_type
constraints
failure_modes_to_avoid… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/hierarchy-delegation-fidelity-under-pressure-v0.1.F1-aero-pressure-coherence-mapping-v0.1What this dataset tests
Whether a system can detectlocalized aero coherence lossfrom pressure sensor fields.
Focus
Platform coherencezone asymmetryvortex integritylocalized collapse zonesbalance shift risk
Required outputs
platform coherence score
zone pressure asymmetry index
vortex system integrity flags
localized collapse zones
balance shift risk score
All scores0 to 1
Higher coherencemeans unified platform.
Higher asymmetry and balance riskmean localized collapse is likely.
clinical-temporal-5node-pressure-buf-lag-cpl-safety-escalation-reg-hold-v0.1
What this repo does
This dataset tests whether a model can detect a safety signal escalation forming over time and predict whether the program crosses into regulatory hold lock-in by the final step.
Core quad
pressurebufferlagcoupling
Prediction target
label_cascade_state
Row structure
One row represents a short temporal window (t0–t3) across program months. It includes time-series values for safety pressure, pharmacovigilance buffer, governance… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-temporal-5node-pressure-buf-lag-cpl-safety-escalation-reg-hold-v0.1.europe-who-mean-systolic-blood-pressure-bp05
Mean systolic blood pressure (crude estimate) | Europe (WHO GHO)
🇪🇺 2,340 observations · 39 Europe countries · 1980–2009 · Repackaged by Electric Sheep Europe
TL;DR
This dataset contains 2,340 observations of Mean systolic blood pressure (crude estimate) data across 39 Europe countries, spanning 1980–2009, covering 1 distinct indicators.
About the source
Source: WHO Global Health Observatory
Publisher: World Health Organization
License:… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepeurope/europe-who-mean-systolic-blood-pressure-bp05.premier-league-linkage-stress-degradation-under-pressure-v0.1What this dataset tests
Whether an intelligence system can estimatehow functional linkages degrade under pressure and fatigueand identify pressure-specific failure points.
Required outputs
stressed linkage loss percent
pressure-specific failure points
fatigue sensitivity index
press exploitability map
time to linkage failure
dominant stressor signature
clinical-quad-endpoint-adjudication-bias-missingness-timing-pressure-v0.1Clarus Clinical Quad Coupling Endpoint Adjudication Bias Missingness Timing Pressure v0.1
What this dataset isThis dataset tests whether a model can detect endpoint adjudication bias created by four interacting forces.
Quad coupling nodes
Endpoint rate or classification shift
Data latency or packet incompleteness
Concomitant exposure or contextual missingness
Governance pressure such as interim look, submission, earnings, or regulator briefing
Input
One vignette
OutputReturn… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-endpoint-adjudication-bias-missingness-timing-pressure-v0.1.clinical-quad-protocol-deviation-cluster-staffing-load-training-gap-governance-pressure-v0.1Clarus Clinical Quad Coupling Protocol Deviation Cluster Staffing Load Training Gap Governance Pressure v0.1
What this dataset isThis dataset tests whether a model can detect clustered protocol deviations caused by four interacting nodes.
Quad coupling nodes
Deviation rate or severity cluster
Staffing or workload pressure
Training gap or outdated materials
Governance or compliance review pressure
Input
One vignette
OutputReturn strict JSON only.
Required output JSON keys… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-quad-protocol-deviation-cluster-staffing-load-training-gap-governance-pressure-v0.1.decision-pressure-mapping-v0.1What this dataset tests
Whether a system can identify non-evidentiary forcesshaping a decision and separate them from evidence.
Required outputs
pressure_sources
pressure_type_map
evidence_vs_pressure_conflicts
pressure_mitigation_actions
Typical failures
mistaking urgency for evidence
authority bias
safety trade-offs hidden by deadlines
Suggested prompt wrapper
System
You map decision pressure and protect evidence integrity.
User
Decision context{decision_context}… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/decision-pressure-mapping-v0.1.infrastructure-temporal-5node-pressure-buf-lag-cpl-grid-stress-blackout-v0.1
What this repo does
This dataset tests whether a model can detect a power grid stress cascade forming over time and predict whether blackout lock-in occurs by the final step.
Core quad
pressurebufferlagcoupling
Prediction target
label_cascade_state
Row structure
One row represents a short time window (t0–t3) of grid stress conditions including demand pressure, reserve buffer margin, response lag, and interconnect coupling tightness. The label marks… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/infrastructure-temporal-5node-pressure-buf-lag-cpl-grid-stress-blackout-v0.1.legal-oral-argument-coherence-pressure-v0.1What this dataset is
You get
a legal issue
a judge question
a counsel answer
context
You decide
Does the answer hold structural coherence under pressure
Task
Answer
coherent
or
incoherent
Only output one word.
What it tests
argument integrity under judicial pressure
concession detection
contradiction detection
structural collapse moments
Why this matters
Most legal analytics looks at outcomes.
This dataset looks at the live failure moment.
The point where an argument breaks in court.
That is… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/legal-oral-argument-coherence-pressure-v0.1.legal-statutory-interpretation-drift-pressure-v0.1What this dataset is
You receive
statutory text
claimed intent
interpretation move
context signals
legislative history
You decide
Does interpretation stay inside text and intent
Answer
coherent
or
incoherent
Why this matters
When text and intent diverge
courts split
appeals rise
amendment pressure builds
This dataset measures statutory coherence decay before doctrinal fracture.
Do full repo ClarusC64/legal-statutory-age-obsolesce
fisheries-market-price-harvest-pressure-coherence-risk-v0.1What this repo is for
Detect when price incentives drive extraction beyond sustainable levels.
Focus
• price spikes
• export demand
• fleet response
• stock levels
This completes the fisheries stack.
You now have:
• biological
• environmental
• governance
• capacity
• economic
clinical-constraint-pressure-v0.2
What this dataset does
This dataset tests whether a model can estimate clinical constraint pressure.
The task is not to identify current illness severity.
The task is to classify how much pressure the patient system is under relative to available reserve.
What changed in v0.2
v0.2 adds counterfactual and adversarial cases.
Some rows have the same oxygen requirement or vasopressor requirement but different reserve states.
Some high-looking cases have preserved… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/clinical-constraint-pressure-v0.2.ai-temporal-5node-pressure-buf-lag-cpl-alignment-goal-drift-v0.1
What this repo does
This dataset tests whether a model can detect an alignment cascade forming over time by reading a short ordered window of signals and predicting whether goal drift lock-in occurs by the final step.
Core quad
pressurebufferlagcoupling
Prediction target
label_cascade_state
Row structure
One row represents one short time window (t0 to t3) for an AI system under alignment pressure. It includes time-series values for optimization… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/ai-temporal-5node-pressure-buf-lag-cpl-alignment-goal-drift-v0.1.