{"data":{"slug":"agi-arena-mars","name":"MARS","tagline":"Advanced optimizer for variance reduction in large model training.","github_url":"https://github.com/AGI-Arena/MARS","owner":"AGI-Arena","repo":"MARS","owner_avatar_url":"https://avatars.githubusercontent.com/u/188802235?v=4","primary_language":"Python","stars":722,"forks":48,"topics":["fine-tuning","large-language-models","optimization-algorithms","optimizer","pretraining"],"archived":false,"github_pushed_at":"2026-03-26T05:08:16+00:00","maintenance_label":"Slowing","stars_delta_30d":-1,"url":"https://www.graphcanon.com/tools/agi-arena-mars","markdown_url":"https://www.graphcanon.com/tools/agi-arena-mars.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/agi-arena-mars","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=agi-arena-mars","description":"The official implementation of MARS: Unleashing the Power of Variance Reduction for Training Large Models","homepage_url":"https://github.com/AGI-Arena/MARS","license":"Apache-2.0","open_issues":14,"watchers":39,"ai_summary":"MARS provides tools and algorithms to enhance the efficiency of training large-scale language models through fine-tuning and pretraining optimization techniques.","readme_excerpt":"### Install Dependencies\n\n```\n$ pip install torch==2.1.2 transformers==4.33.0 datasets tiktoken numpy==1.26.4 wandb\n```","github_created_at":"2024-11-17T22:38:55+00:00","created_at":"2026-07-11T11:40:37.220542+00:00","updated_at":"2026-09-20T06:30:54.879876+00:00","categories":[{"slug":"model-training","name":"Model Training","url":"https://www.graphcanon.com/categories/model-training","markdown_url":"https://www.graphcanon.com/categories/model-training.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/model-training"}],"tags":[{"slug":"fine-tuning","name":"fine-tuning"},{"slug":"large-language-models","name":"large-language-models"},{"slug":"optimization-algorithms","name":"optimization-algorithms"},{"slug":"optimizer","name":"optimizer"},{"slug":"pretraining","name":"pretraining"}],"trust":{"provenance":{"is_fork":false,"github_id":890059746,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-09-19T18:26:45.972Z","maintenance":{"label":"Slowing","score":36,"methodology":"github_public_v1","releases_90d":0,"days_since_push":177,"last_release_at":null,"stars_delta_30d":-1,"open_issues_delta_30d":8},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-11T11:40:38.360Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-09-19T18:26:47.224Z"},"languages":{"value":["python"],"source":"github.language","observed_at":"2026-09-19T18:26:47.224Z"},"license_spdx":{"value":"Apache-2.0","source":"github.license","observed_at":"2026-09-19T18:26:47.224Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["When you need specific tools to reduce variance during the training of large-scale language models","For projects where fine-tuning and pretraining efficiency are critical"],"when_not_to_use":["If your project involves small or medium-sized model training, as MARS is optimized for large-scale scenarios","When other optimization aspects such as memory usage are prioritized over variance reduction"],"source":"enrich:decision_facts","observed_at":"2026-07-15T08:20:50.271Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"MARS focuses on variance reduction for large model training through specialized optimization algorithms."}]}}