{"id":27378778,"url":"https://github.com/reger-men/tensorflow_benchmark","last_synced_at":"2025-04-13T13:37:08.869Z","repository":{"id":36602157,"uuid":"224675132","full_name":"reger-men/tensorflow_benchmark","owner":"reger-men","description":"TensorFlow benchmark scripts for single and multi nodes with multi GPUs","archived":false,"fork":false,"pushed_at":"2023-12-20T10:55:57.000Z","size":66,"stargazers_count":6,"open_issues_count":0,"forks_count":2,"subscribers_count":3,"default_branch":"master","last_synced_at":"2024-11-02T17:12:46.503Z","etag":null,"topics":["multi-gpus","multi-nodes","tensorflow-benchmark","tensorflow2"],"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/reger-men.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}},"created_at":"2019-11-28T14:52:03.000Z","updated_at":"2023-12-20T10:55:54.000Z","dependencies_parsed_at":"2022-08-08T16:01:20.047Z","dependency_job_id":null,"html_url":"https://github.com/reger-men/tensorflow_benchmark","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reger-men%2Ftensorflow_benchmark","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reger-men%2Ftensorflow_benchmark/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reger-men%2Ftensorflow_benchmark/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/reger-men%2Ftensorflow_benchmark/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/reger-men","download_url":"https://codeload.github.com/reger-men/tensorflow_benchmark/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248723243,"owners_count":21151417,"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":["multi-gpus","multi-nodes","tensorflow-benchmark","tensorflow2"],"created_at":"2025-04-13T13:37:08.045Z","updated_at":"2025-04-13T13:37:08.862Z","avatar_url":"https://github.com/reger-men.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Tensorflow v2 benchmark\n![Supports TFv2](https://img.shields.io/badge/Supports-tensorflow%20v2-blue.svg)\n\nTensorFlow benchmark scripts for single and multi Nodes with Multi GPUs\n\n#### Usage\n##### Clone repo\n```git clone https://github.com/reger-men/tensorflow_benchmark.git```\n\n##### Pre-requirement\n```pip3 install -r requirements.txt```\n\n##### Single Node Single GPU\nTrain with custom loop:\n\n```python3 train.py -train_mode loop```\n\nTrain with Keras fit:\n\n```python3 train.py -train_mode fit```\n\n##### Single Node Multi-GPUs\nMirrored strategy will be used as default with ```num_gpus\u003e1```\n\n```python3 train.py -train_mode fit -num_gpus 2```\n\n```python3 train.py -train_mode loop -num_gpus 2```\n\n##### Multi-Node Multi-GPUs\nExperimental: launch Multi-Nodes training from the chief Worker Node\n\n```python3 train.py --train_mode=fit --workers=\"localhost:122,localhost:123\" --w_type=\"worker\" --w_index=0 --distribution_strategy=MultiWorker```\n\nOr with custom loop:\n\n```python3 train.py --train_mode=loop --workers=\"localhost:122,localhost:123\" --w_type=\"worker\" --w_index=0 --distribution_strategy=MultiWorker```\n\n##### Help Flags\n```python3 train.py --helpfull```\n\n```\ntrain.py:\n  --batch_size: Batch Size\n    (default: '128')\n    (an integer)\n  --buffer_size: Shuffle buffer size\n    (default: '50000')\n    (an integer)\n  --display_every: Number of steps after which progress is printed out\n    (default: '20')\n    (an integer)\n  --distribution_strategy: Can be: Mirrored, MultiWorker, OneDevice\n    (default: 'OneDevice')\n  --epochs: Number of epochs\n    (default: '1')\n    (an integer)\n  --num_gpus: Number of GPUs. 0 will run on CPU\n    (default: '1')\n    (an integer)\n  --[no]setup_cluster: Setup the cluster from the chief worker or not. This is an expiremental feature\n    (default: 'true')\n  --train_mode: Use either keras fit or loop training\n    (default: 'fit')\n  --verbose: Set verbosity level\n    (default: '0')\n    (an integer)\n  --w_index: Worker index. 0 is appointed as the chief worker\n    (default: '0')\n    (an integer)\n  --w_type: Task type\n    (default: 'worker')\n  --workers: List of workers IP:Port\n    (default: 'localhost:122,localhost:123')\n  ```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Freger-men%2Ftensorflow_benchmark","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Freger-men%2Ftensorflow_benchmark","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Freger-men%2Ftensorflow_benchmark/lists"}