CONFLICT INTELLIGENCE · SOURCE VS SOURCE

When sources differ, classify the difference before calling it a conflict.

Open Model Weights separates explicit contradictions from scope, granularity and claim-strength differences. Missing evidence never becomes agreement, falsehood or conflict.

0 active conflicts6 comparison signals50/703 current claim-schema coverageMissing stays unknown

ACTIVE SOURCE CONTRADICTIONS

Comparable claims that do not match.

A conflict requires comparable field, scope and claim strength. Parameter naming, component scope and generic license metadata do not automatically qualify.

NO ACTIVE CONFLICTS IN CHECKED COMPARABLE FIELDS

No current source contradiction is retained.

This does not mean every source agrees. It means no contradiction is detected within the currently comparable checked fields; incomplete claim coverage and comparison signals remain explicit.

COMPARISON SIGNALS · NOT CONFLICTS

Differences worth seeing without overstating them.

These values differ, but current evidence does not establish that they describe the same quantity, scope or legal classification. OMW keeps them visible and separate.

01

Comparable values only

Context is compared with context only after current claim parsing; parameter scope and component boundaries remain explicit.

02

Evidence before interpretation

Both source values and URLs remain attached. Classification is a separate OMW layer.

03

No absence inference

A missing model-card claim, license file or runtime document is incomplete evidence, not agreement and not a negative claim.

04

Detector corrections stay visible

When a detector refinement reclassifies an earlier alert, history records the correction rather than pretending it never happened.