MacroMan5MacroMan5/autotrain-yolo/CLAUDE.md

autotrain-yolo

CLAUDE.md8 starsChanged 8 months agoMITView on GitHub

YOLO training toolkit — dataset tools, HP tuning, intelligent experiment automation, CVAT active learning, ONNX export. The /experiment skill runs a 3-step reasoning loop:

Installs packages

In this repository15 other files it gives its agents.
- active-learningSkill
- analyze-resultsSkill
- annotateSkill
- auto-experimentSkill
- benchmarkSkill
- compare-modelsSkill
- cvat-deploySkill
- cvat-pullSkill
- cvat-pushSkill
- explain-resultsSkill
- monitor-trainingSkill
- review-annotationsSkill
- review-datasetSkill
- setup-projectSkill
- training-loopSkill

The file

We split this CLAUDE.md into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Context: Read training-plan.md (boundaries), summary.md (history), dataset_profile.yaml (characteristics)
line 39

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

CLAUDE.md

yolocc

YOLO training toolkit — dataset tools, HP tuning, intelligent experiment automation, CVAT active learning, ONNX export.

Key Files

File	Who edits	Purpose
`training-plan.md`	Human	Training goals, constraints, allowed actions
`yolo-project.yaml`	Human (or /setup)	Project config — classes, dataset, defaults
`experiments/summary.md`	AI	Dashboard of all experiments
`experiments/session_*.md`	AI	Per-session before/after report
`experiments/analysis.md`	AI	Training advisor recommendations
`experiments/dataset_profile.yaml`	AI	Dataset characteristics for architecture selection
`configs/architectures/*.yaml`	Shipped	Pre-built YOLO11 head configs (standard, P2, P2P3P4)
`WORKFLOW.md`	Human	Complete end-to-end workflow guide

Available Slash Commands

/experiment — Run experiment loop (assess → tune → report)
/analyze — Analyze results + write recommendations
/setup — Initialize a new YOLO project (interactive wizard)
/review-dataset — Audit dataset quality + stats
/train — Run a single managed training with report
/review-annotations — AI-assisted annotation review (YOLO + Claude vision)
/annotate — Claude vision annotation correction
/compare-models — Compare 2+ models side-by-side (mAP, per-class AP, speed)
/benchmark — Profile model inference speed and size
/explain-results — Plain-English training results report
/cvat-pull — Pull annotations from CVAT
/cvat-push — Push uncertain images to CVAT for review
/cvat-deploy — Deploy trained model to CVAT via Nuclio
/active-learning — Full active learning loop

How Experiments Work

The /experiment skill runs a 3-step reasoning loop:

Context: Read training-plan.md (boundaries), summary.md (history), dataset_profile.yaml (characteristics)
Reasoning loop (ASSESS → ACT → LOG):
- ASSESS: Classify bottleneck (data quality / architecture mismatch / HP not optimized)
- ACT: Architecture swap, model.tune(), strategic experiment, or data action
- LOG: Read report, append journal entry, decide next step
Guardrails: checkpoint backup, immutable data, architecture change gate
Session report with before/after + key learnings

Common Commands

yolo-train --dataset path/to/dataset --name my_model
yolo-finetune --variant indoor --base models/best.pt
yolo-validate path/to/dataset
yolo-experiment run --override "lr0=0.005" --budget 10
yolo-experiment baseline --budget 5
yolo-experiment summary
yolo-experiment tune --space lr --iterations 20 --epochs 10
yolo-export --model best.pt --imgsz 640
yolo-split --source raw/ --output split/ --classes cat dog bird
yolo-analyze --model best.pt --dataset path/to/dataset
yolo-autolabel --model best.pt --source unlabeled/
yolo-clean --dataset path/to/dataset
yolo-merge --base dataset1/ --overlay dataset2/

Architecture

src/yolocc/
├── project.py          # Config loader (yolo-project.yaml)
├── paths.py            # Workspace path resolution
├── training/           # Train, analyze, utils
├── dataset/            # Validate, split, clean, merge, autolabel
├── export/             # ONNX export
└── experiment/         # Autonomous experimentation engine
    ├── runner.py       # Execute experiments + tune (run_experiment, run_tune)
    ├── tracker.py      # Log + generate reports
    └── strategies.py   # Built-in hyperparameter sweeps (CLI)

For Contributors

Run tests: pytest -v
Install dev: pip install -e ".[dev]"
Lint: ruff check src/

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free