{"id":13738615,"url":"https://github.com/ServiceNow/osaka","last_synced_at":"2025-05-08T16:34:58.668Z","repository":{"id":37974444,"uuid":"246867154","full_name":"ServiceNow/osaka","owner":"ServiceNow","description":"Codebase for \"Online Fast Adaptation and Knowledge Accumulation: a New Approach to Continual Learning\". This is a ServiceNow Research project that was started at Element AI.","archived":false,"fork":false,"pushed_at":"2022-06-27T13:36:57.000Z","size":274,"stargazers_count":103,"open_issues_count":1,"forks_count":10,"subscribers_count":8,"default_branch":"master","last_synced_at":"2024-11-15T07:34:56.358Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ServiceNow.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":"2020-03-12T15:21:34.000Z","updated_at":"2024-11-11T10:01:03.000Z","dependencies_parsed_at":"2022-09-02T13:20:53.535Z","dependency_job_id":null,"html_url":"https://github.com/ServiceNow/osaka","commit_stats":null,"previous_names":["elementai/osaka"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ServiceNow%2Fosaka","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ServiceNow%2Fosaka/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ServiceNow%2Fosaka/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ServiceNow%2Fosaka/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ServiceNow","download_url":"https://codeload.github.com/ServiceNow/osaka/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":253105709,"owners_count":21855084,"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":[],"created_at":"2024-08-03T03:02:29.311Z","updated_at":"2025-05-08T16:34:58.318Z","avatar_url":"https://github.com/ServiceNow.png","language":"Python","readme":"*ServiceNow completed its acquisition of Element AI on January 8, 2021. All references to Element AI in the materials that are part of this project should refer to ServiceNow.*\n\n# Online Fast Adaptation and Knowledge Accumulation:\n# a New Approach to Continual Learning \n\nofficial code for the paper: https://arxiv.org/abs/2003.05856 \n\nNOTE: you can find the version of the repo that can reproduce the paper in the `camera_ready` branch. \u003cbr\u003e\nThe master branch is lagging behind a bit. I'll fast-forward it + clean it up in a bit.\n\n\n## (key) Requirements \n- Python 3.6 or 3.7 (past 3.8 you can't get the proper torch version)\n- Pytorch 1.2 (proper torch version for our torchmeta)\n\n`pip install -r requirements.txt`\n\n## Structure\n\n\n    ├── Config\n        ├── ...                   # model configuration\n    ├── Data\n        ├── omniglot.py           # fetches the dataset      \n        ├── tiered_imagenet.py    # fetches the dataset\n    ├── MAML           \n        ├── metalearner\n            ├── maml.py           # defines the models, in particular their CL strategy\n        ├── model.py              # defines the backbone neural networks\n        ├── utils.py              # some utils    \n    ├── Utils\n        ├── bgd_lib\n            ├── ...         # files for BGD      \n    ├── main.py             # main file\n    ├── dataloaders.py      # defines the experiment setting, constructs the dataloaders    \n    ├── args.py             # arguments\n    ├── template.py         # main file template (if you dont want to use pytorch)\n   \n\n\n## Running Experiments\n\n\nexample: run C-MAML (in verbose mode):\n\n```\npython main.py --model_name ours -v\n```\n\nsome notable args:  \u003c/br\u003e\n\n```\npython main.py --prob_statio 0.98 --num_epochs 0 --cl_strategy always_retrain --meta_lr 0.1 --learn_step_size 1 --per_param_step_size 1 -v \n```\n\nto try different baselines in ['online_sgd', 'fine_tuning', 'MetaCOG', 'MetaBGD', 'MAML','ANIL', 'BGD']  \u003c/br\u003e\n\n`python main.py --model_name \u003cbaseline_name\u003e`\n\n\n## Reproduce the results\n\nSee the [camera-ready](https://github.com/ElementAI/osaka/tree/camera_ready) branch\n\n\n## Logging\n\nLogging is done with [Weights \u0026 Biases](https://www.wandb.com/) and can be turned on like this: \u003c/br\u003e\n`python main.py --wandb \u003cworkspace_name\u003e`\n\n## Reference\n\n```\n@article{caccia2020online,\n  title={Online Fast Adaptation and Knowledge Accumulation: a New Approach to Continual Learning},\n  author={Caccia, Massimo and Rodriguez, Pau and Ostapenko, Oleksiy and Normandin, Fabrice and Lin, Min and Caccia, Lucas and Laradji, Issam and Rish, Irina and Lacoste, Alexandre and Vazquez, David and Charlin, Laurent},\n  journal={NeurIPS},\n  year={2020},\n  url={https://arxiv.org/abs/2003.05856},\n  keywords={Continual-Meta Learning, Setting}\n}\n```\n\n## TODO\n\n- [ ] code to reproduce experiments\n\n## Acknowledgements\n\nMAML code comes from https://github.com/tristandeleu/pytorch-maml\n\n## Contact\n\nmassimo.p.caccia at gmail.com\n\n\n\n\n\n","funding_links":[],"categories":["Python"],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FServiceNow%2Fosaka","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FServiceNow%2Fosaka","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FServiceNow%2Fosaka/lists"}