{"id":13737588,"url":"https://github.com/baiksung/MeTAL","last_synced_at":"2025-05-08T14:33:06.170Z","repository":{"id":38061799,"uuid":"396105399","full_name":"baiksung/MeTAL","owner":"baiksung","description":"Official PyTorch implementation of \"Meta-Learning with Task-Adaptive Loss Function for Few-Shot Learning\" (ICCV2021 Oral)","archived":false,"fork":false,"pushed_at":"2021-12-18T01:41:34.000Z","size":149,"stargazers_count":62,"open_issues_count":4,"forks_count":13,"subscribers_count":2,"default_branch":"master","last_synced_at":"2024-11-15T06:32:06.672Z","etag":null,"topics":["computer-vision","deep-learning","few-shot-learning","iccv","machine-learning","maml","meta-learning","pytorch"],"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/baiksung.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}},"created_at":"2021-08-14T19:15:41.000Z","updated_at":"2024-10-04T14:30:35.000Z","dependencies_parsed_at":"2022-09-03T05:50:27.330Z","dependency_job_id":null,"html_url":"https://github.com/baiksung/MeTAL","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/baiksung%2FMeTAL","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/baiksung%2FMeTAL/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/baiksung%2FMeTAL/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/baiksung%2FMeTAL/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/baiksung","download_url":"https://codeload.github.com/baiksung/MeTAL/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":253085772,"owners_count":21851697,"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":["computer-vision","deep-learning","few-shot-learning","iccv","machine-learning","maml","meta-learning","pytorch"],"created_at":"2024-08-03T03:01:54.227Z","updated_at":"2025-05-08T14:33:05.600Z","avatar_url":"https://github.com/baiksung.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"# MeTAL - Meta-Learning with Task-Adaptive Loss Function for Few-Shot Learning (ICCV2021 Oral)\n#### Sungyong Baik, Janghoon Choi, Heewon Kim, Dohee Cho, Jaesik Min, Kyoung Mu Lee\n\n![](img/MeTAL_overview.png)\n\nOfficial PyTorch implementation of [Meta-Learning with Task-Adaptive Loss Function for Few-Shot Learning (ICCV2021 Oral)](https://openaccess.thecvf.com/content/ICCV2021/papers/Baik_Meta-Learning_With_Task-Adaptive_Loss_Function_for_Few-Shot_Learning_ICCV_2021_paper.pdf)\n\nThe code is based off the public code of [MAML++](https://github.com/AntreasAntoniou/HowToTrainYourMAMLPytorch), where their reimplementation of MAML is used as the baseline.\nThe code also includes the implementation of [ALFA](https://github.com/baiksung/ALFA).\n\n\\[[Paper-arXiv](http://arxiv.org/abs/2110.03909)\\] \\[[Video](https://youtu.be/LHpNSSPsq_Y)\\]\n\n## Requirements\n\nUbuntu 18.04\n- Anaconda3\n- Python==3.7.10\n- PyTorch==1.4\n- numpy==1.19.2\n\nTo install requirements, first download Anaconda3 and then run the following:\n```setup\nconda create -n metal python=3.7.10\nconda activate metal\nbash install.sh\n```\n\n## Datasets\nFor miniIamgenet, the dataset can be downloaded from the [link](https://drive.google.com/file/d/1qQCoGoEJKUCQkk8roncWH7rhPN7aMfBr/view) provided from [MAML++ public code](https://github.com/AntreasAntoniou/HowToTrainYourMAMLPytorch).\nmake a directory named 'datasets' and place the downloaded miniImagnet under the 'datasets' directory.\n\n## Training\n\nTo train a model, run the following command in `experiments_scripts` directory\n\n```bash\nbash MeTAL.sh $GPU_ID\n```\n\n## Evaluation\n\nAfter training is finished, evaluation is performed automatically.\nTo run an evaluation manually, run the same command\n\n```bash\nbash MeTAL.sh $GPU_ID\n```\n\n## Results\n\n|   Model   | Backbone |   1-shot Accuracy   |   5-shot Accuracy   |\n|-----------|----------|---------------------|---------------------|\n|MAML       |  4-CONV  |   49.64 \u0026plusmn; 0.31%   |   64.99 \u0026plusmn; 0.27%   |\n|MeTAL      |  4-CONV  | **52.63 \u0026plusmn; 0.37%** | **70.52 \u0026plusmn; 0.29%** |\n|ALFA+MAML  |  4-CONV  |   50.58 \u0026plusmn; 0.51%   |   69.12 \u0026plusmn; 0.47%   |\n|ALFA+MeTAL |  4-CONV  | **57.75 \u0026plusmn; 0.38%** | **74.10 \u0026plusmn; 0.43%** |\n|MAML       | ResNet12 |   58.60 \u0026plusmn; 0.42%   |   69.54 \u0026plusmn; 0.38%   |\n|MeTAL      | ResNet12 | **59.64 \u0026plusmn; 0.38%** | **76.20 \u0026plusmn; 0.19%** |\n|ALFA+MAML  | ResNet12 |   59.74 \u0026plusmn; 0.49%   |   77.96 \u0026plusmn; 0.41%   |\n|ALFA+MeTAL | ResNet12 | **66.61 \u0026plusmn; 0.28%** | **81.43 \u0026plusmn; 0.29%** |\n\n\n## Reference\n```\n@InProceedings{baik2021meta,\n title={Meta-Learning with Task-Adaptive Loss Function for Few-Shot Learning},\n author={Sungyong Baik, Janghoon Choi, Heewon Kim, Dohee Cho, Jaesik Min, Kyoung Mu Lee}\n booktitle = {International Conference on Computer Vision (ICCV)}, \n year={2021}\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbaiksung%2FMeTAL","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbaiksung%2FMeTAL","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbaiksung%2FMeTAL/lists"}