AgentSkillsCN

apm-ds-baseline

为DS项目构建可复现的基础模型，并记录实验结果。

SKILL.md
--- frontmatter
name: apm-ds-baseline
description: Build a reproducible baseline model for DS projects and record results.
compatibility: opencode

What I do

  • •Establish a representative, domain-credible baseline before experiments.
  • •Ensure the baseline can realistically serve as the foundation for the pipeline.
  • •Enforce explicit hyperparameter choices and logging.

Baseline workflow

  1. •Review memory-bank/ARCHITECTURE.md and EDA findings.
  2. •Define a domain-appropriate baseline model and fixed hyperparameters.
    • •It should be strong enough to compare against, not a toy model.
    • •Prefer a model class that could plausibly remain in the final pipeline.
  3. •Implement baseline in main.py or a standalone script.
  4. •Run a quick validation (or provide commands for the user to run).
  5. •Save artifacts to models/ and logs to logs/.
  6. •Update memory-bank/STATE.md and memory-bank/TASK.md.
  7. •Ensure baseline logs follow apm-logs.

Guardrails

  • •Do not run long training without user approval.
  • •Keep the baseline reproducible and comparable across experiments.