{"id":20663814,"url":"https://github.com/vita-group/scalable-l2o","last_synced_at":"2025-08-21T07:17:10.864Z","repository":{"id":107047175,"uuid":"515244553","full_name":"VITA-Group/Scalable-L2O","owner":"VITA-Group","description":"[ECCV 2022] \"Scalable Learning to Optimize: A Learned Optimizer Can Train Big Models\" by Xuxi Chen, Tianlong Chen, Yu Cheng, Weizhu Chen, Ahmed Awadallah, and Zhangyang Wang","archived":false,"fork":false,"pushed_at":"2022-11-21T03:37:17.000Z","size":384,"stargazers_count":6,"open_issues_count":0,"forks_count":1,"subscribers_count":10,"default_branch":"main","last_synced_at":"2025-06-01T07:12:26.080Z","etag":null,"topics":["learning-to-optimize"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/VITA-Group.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"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,"publiccode":null,"codemeta":null}},"created_at":"2022-07-18T15:40:51.000Z","updated_at":"2023-09-21T05:51:11.000Z","dependencies_parsed_at":"2023-04-13T15:47:12.734Z","dependency_job_id":null,"html_url":"https://github.com/VITA-Group/Scalable-L2O","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/VITA-Group/Scalable-L2O","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/VITA-Group%2FScalable-L2O","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/VITA-Group%2FScalable-L2O/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/VITA-Group%2FScalable-L2O/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/VITA-Group%2FScalable-L2O/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/VITA-Group","download_url":"https://codeload.github.com/VITA-Group/Scalable-L2O/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/VITA-Group%2FScalable-L2O/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":271442253,"owners_count":24760353,"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":"2025-08-21T02:00:08.990Z","response_time":74,"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":["learning-to-optimize"],"created_at":"2024-11-16T19:19:57.986Z","updated_at":"2025-08-21T07:17:10.841Z","avatar_url":"https://github.com/VITA-Group.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Scalable-L2O\n\nImplementation of ECCV 2022 paper: Scalable Learning to Optimize: A Learned Optimizer Can Train Big Models. \n![](framework_l2o.png)\n## Environment\n\nWe recommend using Anaconda to manage the virtual environment. \n\n```bash\nconda env create -f environment.yaml\nconda activate sl2o\n```\n\n## Experiments\n\n### CNNs\n#### Subspaces\n\nWe provide pre-generated subspaces for models in this [link](https://drive.google.com/drive/folders/1PVfLiBXg-n_YknBAlaN8ZpRFiP5B_zdd?usp=sharing). Optionally, one can generate the subspaces by themselves. Please refer to the `subspaces` directory for more details. \n\n#### Meta-Training\n\nResNet-18 (CIFAR-10)\n```bash\npython -u src/resnet18_ft_de.py --max_epoch 20 --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 128 --unroll 10 --meta_train_eval_epoch 2 \n```\n\nResNet-18 (CIFAR-100)\n```bash\npython -u src/resnet18_ft_de.py --max_epoch 20 --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 128 --unroll 10 --meta_train_eval_epoch 2 --dataset CIFAR100\n```\n\n\nResNet8 (CIFAR-10)\n```bash\npython -u src/resnet8_ft_de.py --max_epoch 20 --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 128 --unroll 10 --meta_train_eval_epoch 2 \n```\n\nResNet20 (CIFAR-10)\n```bash\npython -u src/resnet20_ft_de.py --max_epoch 20 --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 128 --unroll 10 --meta_train_eval_epoch 2\n```\n\n#### Meta-Testing\n\nResNet-8 (CIFAR-10)\n```bash\npython -u src/resnet8_eval_de.py --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 128 --unroll 10 --max_epoch 100\n```\n\nResNet-20 (CIFAR-10)\n```bash\npython -u src/resnet20_eval_de.py --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 128 --unroll 10 --max_epoch 100\n```\n\n### VITs\n\n#### Meta-Training\n```bash\npython -u src/vit_ft.py --max_epoch 20 --lora_dim 16 --lora_alpha 32 --lora_dropout 0.1 --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 5 --training_steps 1000 --batch-size 64 --unroll 10 --random_seed 1 --name cifar10-100_500 --dataset cifar10 --model_type ViT-B_16 --pretrained_dir checkpoint/ViT-B_16.npz --meta_train_eval_epoch 2 \n```\n\n#### Meta-Testing\n```bash\npython -u src/vit_ft_eval.py --max_epoch 20 --lora_dim 16 --lora_alpha 32 --lora_dropout 0.1 --eval_interval 2000 --log_interval 100 --hidden_sz 8 --scale 1e-4 --log_interval 10 --training_steps 1000 --batch-size 64 --unroll 10 --random_seed 1 --name cifar10-100_500 --dataset cifar10 --model_type ViT-B_16 --pretrained_dir checkpoint/ViT-B_16.npz --random_seed 1 --eval_interval 391 \n```","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fvita-group%2Fscalable-l2o","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fvita-group%2Fscalable-l2o","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fvita-group%2Fscalable-l2o/lists"}