{"id":28429079,"url":"https://github.com/mindsdb/open-sigmoid","last_synced_at":"2025-07-29T05:09:06.496Z","repository":{"id":230091156,"uuid":"774329576","full_name":"mindsdb/open-sigmoid","owner":"mindsdb","description":"Open Source codebase of SIGMOID, the Scalable Infrastructure for Generic Model Optimization on Inhomogeneous Datasets","archived":false,"fork":false,"pushed_at":"2024-03-22T20:22:47.000Z","size":189,"stargazers_count":9,"open_issues_count":0,"forks_count":3,"subscribers_count":13,"default_branch":"staging","last_synced_at":"2025-07-18T10:48:42.692Z","etag":null,"topics":["automl","machine-learning"],"latest_commit_sha":null,"homepage":"","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/mindsdb.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"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}},"created_at":"2024-03-19T11:08:39.000Z","updated_at":"2024-04-05T07:13:59.000Z","dependencies_parsed_at":"2024-03-27T20:24:37.566Z","dependency_job_id":"cbd67813-8e04-448d-ac16-f383a0327c25","html_url":"https://github.com/mindsdb/open-sigmoid","commit_stats":null,"previous_names":["mindsdb/open-sigmoid"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/mindsdb/open-sigmoid","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mindsdb%2Fopen-sigmoid","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mindsdb%2Fopen-sigmoid/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mindsdb%2Fopen-sigmoid/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mindsdb%2Fopen-sigmoid/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mindsdb","download_url":"https://codeload.github.com/mindsdb/open-sigmoid/tar.gz/refs/heads/staging","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mindsdb%2Fopen-sigmoid/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":267633101,"owners_count":24118752,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-07-29T02:00:12.549Z","response_time":2574,"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":["automl","machine-learning"],"created_at":"2025-06-05T13:08:32.140Z","updated_at":"2025-07-29T05:09:06.489Z","avatar_url":"https://github.com/mindsdb.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# open-sigmoid\n\nOpen Source codebase of SIGMOID, the Scalable Infrastructure for Generic Model Optimization on Inhomogeneous Datasets.\n\n## Description\n\n**SIGMOID** stands for **S**calable  **I**nfrastructure for **G**eneric **M**odel  **O**ptimization on **I**nhomogeneous  **D**atasets. It is an infrastructure in the sense that is is not a _single_ computer program but rather a _collection_ of them. The main goal of `sigmoid` is to provide scalabitility to an already existing model. In short, this means\n\n- Making it possible to train a arbitrary model using as much data as possible without changing the model at all.\n- Provide the output product in a form-factor that suits large-scale HPC compute infrastructure.\n- Accomplish the above with zero Human intervention.\n\n## High-level overview\n\n### Data-driven model scaling\n\nA key distinction between `sigmoid` and already existing solutions is that `sigmoid` relies on the training data itself to provide scalability. We call this method \"data-driven model scaling\" (D2MS).\n\n`sigmoid` attempts to achieve D2MS by combining self-supervised Deep Learning methods and unsupervised clustering algorithms to detect underlying data partitions in the dataset; loosely speaking, a partition is a subset of the data where every all elements are similar to one another.\n\n`sigmoid` then trains an arbitrary number of models in a way that makes every model become specialized (fine-tuned) for data coming from one particular partition. This way, no instance of the model gets to \"see\" the entire dataset.\n\nFinally, after the training process, `sigmoid` provides the user with a \"pool\" of models (the specialists) and a \"routing\" model (a switch). Inference then comes down to feeding new data to the switch, which redirects the data to the respective specialist to perform the actual inference.\n\n![High level flow-diagram of `sigmoid`](/assets/figures/sigmoid_flow_diagram.png)\n\n## Installation\n\n`sigmoid` is written in Python, so to install it from source need a Python Environment (recommended to use `pyenv`) and `poetry`.\n\n```:shell\npip install poetry\npoetry install --only main\n```","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmindsdb%2Fopen-sigmoid","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmindsdb%2Fopen-sigmoid","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmindsdb%2Fopen-sigmoid/lists"}