Menu
Log ML experiments with hyperparameters, metrics, and plots; human interprets results and plans next experiments
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
| name | experiment-logger |
| description | Log ML experiments with hyperparameters, metrics, and plots; human interprets results and plans next experiments |
You are executing the experiment-logger skill.
$ARGUMENTS
Track ML experiments systematically. AI logs and visualizes; you interpret and decide next steps.
| Phase | Actor | Action |
|---|---|---|
| 1 | Human | Define experiment goals |
| 2 | Coder | Set up logging (params, metrics) |
| 3 | Coder | Run experiment |
| 4 | Coder | Generate plots (loss curves, comparisons) |
| 5 | Writer | Summarize results |
| 6 | Human | Interpret, decide next experiments |
experiments/
├── exp_001_baseline/
│ ├── config.yaml
│ ├── metrics.json
│ ├── plots/
│ │ ├── loss.png
│ │ └── accuracy.png
│ └── summary.md
├── exp_002_lr_sweep/
│ └── ...
└── comparison.md
# Experiment: [name]
## Config
- learning_rate: 0.001
- batch_size: 32
- epochs: 100
## Results
| Metric | Value | vs Baseline |
|--------|-------|-------------|
| Loss | 0.23 | -15% |
| Acc | 94.2% | +2.1% |
## Observations
- Converged faster than baseline
- Some overfitting after epoch 80
## Plots
Use when user wants to sync, push, or add BibTeX entries to a Zotero collection/folder. Handles deduplication by DOI, arXiv ID, or title.
Update skills with concise concrete practices, keeping SKILL.md minimal
Evaluate necessity and difficulty of issues, polish body, and create on GitHub with AI-assisted triage
Research academic papers and questions, write typst documents with cetz visualizations, teach until human understands
Autonomous daily digest of relevant arxiv papers in physics and ML based on your interests
Find, format, and manage academic citations in BibTeX and typst formats