L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Productivité · SKILL.md

rectified_ode_reflow

rectified_ode_reflow

Simulate rectified-flow Euler transport and compute straightness, reflow, and convex transport diagnostics.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: rectified_ode_reflow description: Simulate rectified-flow Euler transport and compute straightness, reflow, and convex transport diagnostics.

Rectified ODE Reflow

Use this skill when a learned rectified-flow velocity field must be integrated or audited for straightness. Do not use it to fit the velocity field.

Inputs

  • Initial samples z0.
  • Affine velocity parameters compatible with rectified_velocity_regression.
  • Euler step count.

Outputs

  • Final transported samples.
  • Path length and direct endpoint distance.
  • Straightness ratio and squared transport cost.

Workflow

  1. Integrate z <- z + dt * v(z,t) using Euler steps.
  2. Accumulate segment lengths for each sample.
  3. Compare accumulated path length with direct endpoint distance.
  4. Report one-step and multi-step diagnostics for reduced recovery.

Validation

Run python tests/test_ode.py or the Distiller validator with tests enabled.

Limitations

The script checks the ODE/reflow mechanism on numeric vectors. Full image quality metrics require a separate heavy runtime that is outside this bounded reduced recovery.

Skills enregistrées