{"id":48219269,"url":"https://github.com/apple/ml-ssd","last_synced_at":"2026-04-06T21:00:45.684Z","repository":{"id":349048285,"uuid":"1196596858","full_name":"apple/ml-ssd","owner":"apple","description":null,"archived":false,"fork":false,"pushed_at":"2026-04-03T23:38:05.000Z","size":335,"stargazers_count":416,"open_issues_count":2,"forks_count":35,"subscribers_count":7,"default_branch":"main","last_synced_at":"2026-04-05T20:02:26.594Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/apple.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"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":"2026-03-30T21:14:58.000Z","updated_at":"2026-04-05T19:55:10.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/apple/ml-ssd","commit_stats":null,"previous_names":["apple/ml-ssd"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/apple/ml-ssd","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apple%2Fml-ssd","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apple%2Fml-ssd/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apple%2Fml-ssd/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apple%2Fml-ssd/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/apple","download_url":"https://codeload.github.com/apple/ml-ssd/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apple%2Fml-ssd/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31489427,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-06T17:22:55.647Z","status":"ssl_error","status_checked_at":"2026-04-06T17:22:54.741Z","response_time":112,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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":[],"created_at":"2026-04-04T19:03:45.206Z","updated_at":"2026-04-06T21:00:45.665Z","avatar_url":"https://github.com/apple.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"# Simple Self-Distillation\n\u003cdiv align=\"center\"\u003e\n\n[![arXiv](https://img.shields.io/badge/arXiv-2604.01193-b31b1b.svg)](https://arxiv.org/abs/2604.01193)\n[![License](https://img.shields.io/badge/License-Apple-blue)](LICENSE)\n[![Python](https://img.shields.io/badge/Python-3.10+-green.svg)](https://www.python.org/)\n\n### Embarrassingly Simple Self-Distillation Improves Code Generation\n\nRuixiang Zhang\\*, Richard He Bai\\*, Huangjie Zheng\\*, Navdeep Jaitly, Ronan Collobert, Yizhe Zhang\\*\n\n\u003csub\u003e\\*Equal contribution\u003c/sub\u003e\n\n\u003c/div\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"figures/fig_teaser.png\" width=\"100%\" alt=\"SSD Overview\"\u003e\n\u003c/p\u003e\n\n## ✨ Overview\n\nThis repository reproduces the method from the paper:\n\n\u003e **Embarrassingly Simple Self-Distillation Improves Code Generation**\n\nThe approach consists of three simple steps:\n\n1. **Sample** solutions from a frozen model at non-unit temperature  \n2. **Fine-tune** on raw, unverified outputs using standard cross-entropy  \n3. **Decode** with a separately tuned temperature  \n\n**No rewards · No verifier · No teacher · No RL**\n\nFor full details, see the [paper](https://arxiv.org/abs/2604.01193).\n\n---\n\n## 📰 News\n\n- **[2026-04-03]** 🚀 Initial release of repository  \n- **[2026-04-03]** 🤗 Model checkpoints coming soon on Hugging Face  \n- *(More updates will be added here)*\n\n---\n\n## 🚀 Getting Started\n\n```bash\ngit clone https://github.com/apple/ml-ssd.git\ncd ml-ssd\nuv sync --group evaluation\n```\n\n\n\u003cdetails\u003e\n\u003csummary\u003eEvaluation commands\u003c/summary\u003e\n\n```bash\nsource .venv/bin/activate\npython evaluation/eval.py \\\n    --model \u003chf_model_name\u003e \\\n    --tensor_parallel_size 4 \\\n    --max_tokens 65536 \\\n    --n_repeat 10 \\\n    --sampling_params \"temperature=0.9,top_p=0.8,top_k=20\" \\\n    --output_path ./results/\n```\n\n\u003e **Note:** The sampling parameters above are illustrative. Please refer to each model's HuggingFace model card for the recommended sampling parameters.\n\n\u003c/details\u003e\n\n## 🤗 Models\n\u003e Note: Model checkpoints are coming soon. Stay tuned!\n\n## 📁 Repository Structure\n\n```\n├── evaluation/\n│   ├── eval.py                  # CLI entry point\n│   ├── benchmark.py             # LiveCodeBench v6 implementation\n│   └── livecodebench_utils.py   # Code execution utilities\n├── figures/\n│   └── fig_teaser.png\n├── pyproject.toml\n└── README.md\n```\n\n## 📝 Citation\n\n```bibtex\n@misc{zhang2026embarrassinglysimpleselfdistillationimproves,\n      title={Embarrassingly Simple Self-Distillation Improves Code Generation},\n      author={Ruixiang Zhang and Richard He Bai and Huangjie Zheng and Navdeep Jaitly and Ronan Collobert and Yizhe Zhang},\n      year={2026},\n      eprint={2604.01193},\n      archivePrefix={arXiv},\n      primaryClass={cs.CL},\n      url={https://arxiv.org/abs/2604.01193},\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapple%2Fml-ssd","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fapple%2Fml-ssd","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapple%2Fml-ssd/lists"}