Paper to Pipeline

根据机器学习/深度学习论文的实验规划文档自动生成完整的 Python 实验 pipeline。支持数据预处理、模型构建、训练循环、评估指标、结果可视化。Use when user uploads an experiment plan document and wants to generate runnable...

This skill is plausible but inconsistent and may not deliver fully runnable experiments out of the box. Before installing or running: 1) Note missing files: SKILL.md references several templates and a training-best-practices doc that are not in the package — expect gaps. 2) Inspect scripts/generate_pipeline.py and any generated files before executing: look for network calls (e.g., transformers.from_pretrained, model weight downloads) and any filesystem/network operations you did not expect. 3) Run generated code in an isolated/sandbox environment (no sensitive data, limited network) to confirm behavior. 4) Be prepared to manually fill template TODOs (DataLoader, dataset paths, dependency list). 5) If you want higher assurance, provide the full generate_pipeline.py (complete file) and any missing templates so a deeper review can confirm there is no hidden I/O or unexpected external endpoints.

Type: OpenClaw Skill Name: paper-to-pipeline Version: 1.0.0 The 'paper-to-pipeline' skill bundle is a legitimate tool designed to automate the generation of machine learning experiment pipelines from structured markdown plans. The core logic in 'scripts/generate_pipeline.py' uses regex to parse user-provided text and writes standard PyTorch/Transformers boilerplate code to a specified directory. No evidence of data exfiltration, malicious execution, or prompt injection was found; the templates and scripts follow standard ML development practices.