{"id":23219225,"url":"https://github.com/shenggan/atp","last_synced_at":"2025-08-19T09:32:30.514Z","repository":{"id":102455776,"uuid":"559809562","full_name":"Shenggan/atp","owner":"Shenggan","description":"Adaptive Tensor Parallelism for Foundation Models","archived":false,"fork":false,"pushed_at":"2022-12-15T11:17:17.000Z","size":3373,"stargazers_count":7,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2024-01-29T20:34:33.976Z","etag":null,"topics":["attention","distributed-training","gpt","large-model","model-parallelism","pytorch","transformer"],"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/Shenggan.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}},"created_at":"2022-10-31T06:15:27.000Z","updated_at":"2023-08-11T10:58:02.000Z","dependencies_parsed_at":"2023-09-25T06:57:19.715Z","dependency_job_id":null,"html_url":"https://github.com/Shenggan/atp","commit_stats":{"total_commits":16,"total_committers":2,"mean_commits":8.0,"dds":0.0625,"last_synced_commit":"5edf1bd8e0f2f7759e192da14344630fd31583c5"},"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Shenggan%2Fatp","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Shenggan%2Fatp/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Shenggan%2Fatp/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Shenggan%2Fatp/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Shenggan","download_url":"https://codeload.github.com/Shenggan/atp/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":230342834,"owners_count":18211594,"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","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":["attention","distributed-training","gpt","large-model","model-parallelism","pytorch","transformer"],"created_at":"2024-12-18T21:30:46.545Z","updated_at":"2024-12-18T21:30:47.074Z","avatar_url":"https://github.com/Shenggan.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"![](./assert/banner.png)\n\n# ATP: Adaptive Tensor Parallelism\n\nAdaptive Tensor Parallelism for Large Model Traning and Inference\n\nATP provides a high-performance implementation of Topology-aware Tensor Parallelism with the following characteristics.\n\n1. Two-Level Search Space for Tensor Parallelism.\n2. Adaptive Tensor Parallelism with Hierarchical Communication Matrix.\n3. Chunk-based Communication-Computation Overlapping.\n4. An estimator that helps study the performance of ATP on networks with different topologies.\n\n## Installation\n\nTo install ATP, you will need:\n\n+ Python 3.8 or 3.9.\n+ PyTorch 1.13\n+ SPMD from [pytorch/tau](https://github.com/pytorch/tau)\n\n```\nconda install pytorch torchvision torchaudio pytorch-cuda=11.6 -c pytorch -c nvidia\npip install git+https://github.com/pytorch/tau.git@89700fd\n```\n\n## Usage\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fshenggan%2Fatp","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fshenggan%2Fatp","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fshenggan%2Fatp/lists"}