ClarusC64/euv-stochastic-yield-failure-horizon-routing-v0.1
Purpose Yield does not collapse instantly.It drifts. The early signal is coherence decay between: source noiseresist responseLER spreaddefect clusteringyield loss This dataset predicts: how far the system driftedhow many lots remain before yield collapsewhat intervention stabilizes the line Task Return JSON: drift_scorefailure_horizon_lotsintervention_route Example {"drift_score":0.58,"failure_horizon_lots":60,"intervention_route":"laser_tuning"}… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/euv-stochastic-yield-failure-horizon-routing-v0.1.
Purpose
Yield does not collapse instantly. It drifts.
The early signal is coherence decay between:
source noise resist response LER spread defect clustering yield loss
This dataset predicts:
how far the system drifted how many lots remain before yield collapse what intervention stabilizes the line
Task
Return JSON:
driftscore failurehorizonlots interventionroute
Example
{"driftscore":0.58,"failurehorizonlots":60,"interventionroute":"laser_tuning"}
Route options
none monitor resistrecalibration sourcestabilization lasertuning coupledadjustment stopandclean halt_process
Inputs
All values are deltas vs a known stable window.
photonfluxcvdelta sourcepulseenergycvdelta resistsensitivitydeltapct lersigmadeltanm stochasticdefectppmdelta yielddeltapct coherencescoredelta
Output meaning
drift_score Distance from stable process manifold
failurehorizonlots Lots remaining before yield becomes non-viable
intervention_route Minimal action to restore coherence
Version
v0.1