datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
long-horizon-drift-v0.1c
What this dataset tests
Long arcs bend.
Past stability can hide future risk.
Why it exists
Long plans fail when drift accumulates.
This set checks whether you
detect slow shifts
reject frozen baselines
project forward risk
set milestones
Data format
Each row contains
long_horizon_context
user_message
drift_pressure
constraints
failure_modes_to_avoid
target_behaviors
gold_checklist
Feed the model
long_horizon_context
user_message… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/long-horizon-drift-v0.1c.mid-horizon-drift-v0.1b
What this dataset tests
Mid-range trends can lie.
Mix shifts.
Definitions shift.
Leading indicators speak first.
Why it exists
Models over-trust blended mid-horizon metrics.
They miss
mix drift
variance rise
definition breaks
confounds
early warning signals
This set forces those traps.
Data format
Each row contains
mid_horizon_context
user_message
drift_pressure
constraints
failure_modes_to_avoid
target_behaviors
gold_checklist
Feed the… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/mid-horizon-drift-v0.1b.ai-goal-failure-horizon-and-realignment-routing-v0.1What this dataset is
Predicts how soon goal drift becomes a hard failure
Names the realignment window before collapse
Forces an intervention choice with triggers and monitoring
Inputs
setting
env_shift_event
observed_drift_markers
goal_representation_summary
behavioral_deviation_summary
system_constraints
intervention_options
Gold fields in the CSV
failure_mode
estimated_failure_horizon_steps
realignment_window_steps
gold_intervention_choice
realignment_trigger_conditions… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/ai-goal-failure-horizon-and-realignment-routing-v0.1.