ai-research-reproduction
Reproduce deep learning code with run evidence.
What it does
Guides a minimal, README-grounded deep learning reproduction run and produces an auditable evidence bundle.
When to use it
Use for repository-grounded, multi-phase reproduction aimed at a small reproducible target, not standalone commands, generic setup, or paper summaries.
How to use it
Give it a deep learning repository to reproduce; it runs a bounded documented target and writes the reproduction evidence bundle.
What you provide
- An existing project
What you need · 5
Requires Python 3.11 or newer for bundled orchestration; additional target-repository dependencies must be reviewed.
Routes inference, evaluation, and smoke runs to this skill; the files do not state that it must be installed separately.
Routes training startup, kickoff, or resume to this skill when training is selected; a separate installation requirement is unstated.
Access · 1
This skill
host CPU and memory
Reads host CPU and memory