The interface is localized. Skill descriptions and documents are machine-translated by AI into your language; code, commands, paths and technical terms stay in the original, and the author's document governs.
Explore skills
Development · SKILL.md

sr3_iterative_sampler

sr3_iterative_sampler

Apply the SR3 reverse-chain update from Gaussian noise to a refined high-resolution image. This reusable skill supports bounded SR3 recovery experiments.

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill documentation

GitHub

name: sr3_iterative_sampler description: Apply the SR3 reverse-chain update from Gaussian noise to a refined high-resolution image. This reusable skill supports bounded SR3 recovery experiments.

sr3_iterative_sampler

Use this skill for SR3 proxy recovery when implementing iterative_sampler. Inputs are tiny numeric arrays or metadata; outputs are deterministic logs or metrics. Workflow follows SR3 paper equations for degradation, noising, denoising loss, reverse refinement, or evaluation as appropriate. Validate with validate_skill_tree.py --run-tests. Limitations: not a full SR3 U-Net implementation and not a substitute for FFHQ/ImageNet-scale training.

Saved skills