---

name: checkpoint-tools description: "Extract, compress, validate, and compose Custom Diffusion checkpoint deltas." disable-model-invocation: true metadata: disco-role: operating license: NOASSERTION

Checkpoint tools

Use this sub-skill for delta and checkpoint math only.

It covers:

• legacy checkpoint delta extraction
• low-rank compression of K/V deltas
• layout validation for extracted, compressed, and composed deltas
• diffusers-side concept composition outputs

It does not cover image generation after composition; route that work to ../inference/SKILL.md.

Start here

1. Read references/api-reference.md.
2. Use scripts/check_delta_layout.py before sampling or recompressing.
3. Read references/workflows.md for extraction, compression, and composition order.
4. Check references/troubleshooting.md when a checkpoint family or tensor layout is wrong.

Safe defaults

• Extraction is CPU-safe and does not delete anything unless --delete-source is set.
• Compression and composition are expected to use CUDA and local weights for real runs.
• Layout checks are CPU-safe and do not download models.

Preserve the upstream Adobe Research notice if you edit the bundled helper scripts.