{"id":48695494,"url":"https://github.com/omnipotence-eth/ml-lab","last_synced_at":"2026-04-11T07:13:36.193Z","repository":{"id":350231063,"uuid":"1205908781","full_name":"omnipotence-eth/ml-lab","owner":"omnipotence-eth","description":"ML research control plane — experiment lifecycle, model registry, cloud training launcher","archived":false,"fork":false,"pushed_at":"2026-04-09T11:58:38.000Z","size":40,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-04-09T13:29:36.669Z","etag":null,"topics":["cloud-training","experiment-tracking","gpu-diagnostics","ml","mlops","model-registry","python","pytorch","research","wandb"],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/omnipotence-eth.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":"SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-04-09T11:54:56.000Z","updated_at":"2026-04-09T11:58:36.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/omnipotence-eth/ml-lab","commit_stats":null,"previous_names":["omnipotence-eth/ml-lab"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/omnipotence-eth/ml-lab","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/omnipotence-eth%2Fml-lab","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/omnipotence-eth%2Fml-lab/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/omnipotence-eth%2Fml-lab/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/omnipotence-eth%2Fml-lab/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/omnipotence-eth","download_url":"https://codeload.github.com/omnipotence-eth/ml-lab/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/omnipotence-eth%2Fml-lab/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31671999,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-10T17:19:37.612Z","status":"online","status_checked_at":"2026-04-11T02:00:05.776Z","response_time":54,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["cloud-training","experiment-tracking","gpu-diagnostics","ml","mlops","model-registry","python","pytorch","research","wandb"],"created_at":"2026-04-11T07:13:35.560Z","updated_at":"2026-04-11T07:13:36.187Z","avatar_url":"https://github.com/omnipotence-eth.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n\u003ch1\u003eml-lab\u003c/h1\u003e\n\n\u003cp\u003eML research control plane — experiment lifecycle, model registry, cloud training launcher\u003c/p\u003e\n\n![CI](https://img.shields.io/github/actions/workflow/status/omnipotence-eth/ml-lab/ci.yml?style=flat-square\u0026label=CI)\n![Python](https://img.shields.io/badge/Python-3.11+-blue?style=flat-square)\n![W\u0026B](https://img.shields.io/badge/W%26B-integrated-yellow?style=flat-square)\n![Ruff](https://img.shields.io/badge/code%20style-ruff-000000?style=flat-square)\n![License](https://img.shields.io/badge/License-MIT-green?style=flat-square)\n\n\u003c/div\u003e\n\n\u003e Demo coming soon\n\n## Why\n\nRunning ML experiments across local hardware and cloud GPUs produces scattered checkpoints, siloed W\u0026B projects, and no systematic way to compare results. ml-lab connects existing tools (ml-experiment-scaffold, gpu-server-test-suite, llm-wiki) into a unified 7-stage lifecycle: preflight → init → configure → train → eval → register → publish. Same configs work locally on an RTX 5070 Ti and on cloud A100s.\n\n## Features\n\n- **Experiment initialization** from ml-experiment-scaffold templates\n- **GPU preflight checks** via gpu-server-test-suite before training\n- **Config validation** catches impossible hyperparameter combos (fp8 training, OOM configs)\n- **Cloud training** with rsync + SSH to RunPod, Lambda, or vast.ai\n- **Model registry** — append-only JSONL with eval scores, config hashes, metadata\n- **Cross-experiment leaderboard** for comparing models across methods and seeds\n- **Automated W\u0026B sync** for Device Guard environments via WSL\n- **Knowledge integration** — publish findings to llm-wiki\n\n## Architecture\n\n```mermaid\ngraph TD\n    ML[ml-lab\u003cbr/\u003eControl Plane] --\u003e SC[ml-experiment-scaffold\u003cbr/\u003eTemplate]\n    ML --\u003e GPU[gpu-server-test-suite\u003cbr/\u003ePreflight]\n    ML --\u003e WIKI[llm-wiki\u003cbr/\u003eKnowledge Base]\n\n    subgraph \"Experiment Lifecycle\"\n        P[1. Preflight] --\u003e I[2. Init]\n        I --\u003e C[3. Configure]\n        C --\u003e T[4. Train]\n        T --\u003e E[5. Eval]\n        E --\u003e R[6. Register]\n        R --\u003e PB[7. Publish]\n    end\n\n    ML --\u003e P\n\n    subgraph \"Training Targets\"\n        LOCAL[Local RTX 5070 Ti]\n        CLOUD[Cloud A100/H100]\n    end\n\n    T --\u003e LOCAL\n    T --\u003e CLOUD\n```\n\n## Quick Start\n\n```bash\n# Clone\ngit clone https://github.com/omnipotence-eth/ml-lab.git\ncd ml-lab\n\n# Install\npip install -e \".[dev]\"\n\n# Create a new experiment\nmake new-experiment NAME=gsm8k-grpo\n\n# Validate config\nmake validate-config EXP=2026-04-gsm8k-grpo\n\n# Run preflight + train\nmake train EXP=2026-04-gsm8k-grpo\n\n# Register model after training\nmake register EXP=2026-04-gsm8k-grpo\n\n# View leaderboard\nmake leaderboard\n```\n\n## Project Structure\n\n```\nml-lab/\n├── experiments/          # Experiment instances (from scaffold template)\n│   └── YYYY-MM-\u003cname\u003e/  # Each experiment with configs, src, results\n├── registry/\n│   ├── models.jsonl      # Append-only model index\n│   └── README.md         # Schema documentation\n├── cloud/\n│   ├── providers.yaml    # RunPod/Lambda/vast.ai configs\n│   ├── launch.py         # rsync + SSH orchestrator\n│   ├── Dockerfile.train  # Training container\n│   └── setup_remote.sh   # One-shot remote env setup\n├── scripts/\n│   ├── new_experiment.py # Init from scaffold template\n│   ├── preflight.py      # GPU health check\n│   ├── register_model.py # Post-training registration\n│   ├── cross_compare.py  # Leaderboard generator\n│   ├── sync_wandb.py     # WSL-based W\u0026B sync\n│   └── research_to_wiki.py # Push findings to llm-wiki\n├── src/ml_lab/\n│   ├── cli.py            # Click CLI\n│   └── config_validator.py # Config validation\n├── tests/                # pytest test suite\n├── Makefile              # Top-level orchestration\n└── pyproject.toml\n```\n\n## Development\n\n```bash\n# Run tests\nmake test\n\n# Lint + format\nmake lint\n\n# Run specific test\npytest tests/test_config_validator.py -v\n```\n\n## License\n\n[MIT](LICENSE)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fomnipotence-eth%2Fml-lab","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fomnipotence-eth%2Fml-lab","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fomnipotence-eth%2Fml-lab/lists"}