{"id":13628006,"url":"https://github.com/mingzhang96/MAS-mxnet","last_synced_at":"2025-04-17T00:32:59.275Z","repository":{"id":217076584,"uuid":"161596823","full_name":"mingzhang96/MAS-mxnet","owner":"mingzhang96","description":"This is an implement of MAS-Memory-Aware-Synapses on MXNet.","archived":false,"fork":false,"pushed_at":"2019-03-11T06:46:09.000Z","size":15,"stargazers_count":4,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-04-01T21:07:06.726Z","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":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/mingzhang96.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":"2018-12-13T06:54:40.000Z","updated_at":"2022-11-05T11:12:34.000Z","dependencies_parsed_at":null,"dependency_job_id":"b4e09e8c-9800-4676-b5ba-d765b8b03e02","html_url":"https://github.com/mingzhang96/MAS-mxnet","commit_stats":null,"previous_names":["mingzhang96/mas-mxnet"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mingzhang96%2FMAS-mxnet","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mingzhang96%2FMAS-mxnet/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mingzhang96%2FMAS-mxnet/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mingzhang96%2FMAS-mxnet/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mingzhang96","download_url":"https://codeload.github.com/mingzhang96/MAS-mxnet/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":249293149,"owners_count":21245686,"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":"2024-08-01T22:00:42.679Z","updated_at":"2025-04-17T00:32:59.029Z","avatar_url":"https://github.com/mingzhang96.png","language":"Python","funding_links":[],"categories":["\u003ca name=\"Vision\"\u003e\u003c/a\u003e2. Vision"],"sub_categories":["2.14 Misc"],"readme":"## Implement of MAS on MXNet\n\nThis is an implement of MAS on MXNet. \n\n[Origin MAS on pytorch](https://github.com/rahafaljundi/MAS-Memory-Aware-Synapses)\n\n## what does this project finish\n\n* standard setup and training on several task.\n* finally calculate accuracy on each task.\n\n## environment\n* mxnet-cu80 on version 1.1.0.post0\n* python 2.7\n\n## how to use\n\n1. clone the project\n```shell\n$ git clone https://github.com/mingzhang96/MAS-mxnet.git\n$ cd MAS-mxnet\n$ mkdir ckpt \u0026\u0026 mkdir data \u0026\u0026 mkdir reg_params\n```\n\n2. We assume that you are in the `$MAS-mxnet` directory, and in `$MAS-mxnet/data` the mnist (`.gz`) data stays there.\n```shell\npython train_mnist.py\n```\n\n## result\n\n*we use mlp instead of AlexNet as our base model.*\n\n**notice: we use model trained on last task to test other tasks.**\n\n#### 100 epoch, update_lr = 0.05, train_lr = 0.05\n\ntask | accuracy\n---|---\n01 | 0.6274231678486998\n23 | 0.9417238001958864\n45 | 0.9797225186766275\n67 | 0.972306143001007\n89 | 0.9389813414019162\n\n#### 200 epoch, update_lr = 0.0001, train_lr = 0.0008\n\ntask | accuracy\n---|---\n01 | 0.9952718676122931\n23 | 0.8805093046033301\n45 | 0.955709711846318\n67 | 0.9823766364551864\n89 | 0.9536056480080686\n\n#### 200 epoch, update_lr = 0.0001, train_lr = 0.005, fc2.output = 256\n\ntask | accuracy\n---|---\n01 | 0.9933806146572104\n23 | 0.9299706170421156\n45 | 0.9802561366061899\n67 | 0.9914400805639476\n89 | 0.9646999495713565\n\n## tips\n\n* the more tasks are, the more epoch need to train.\n* use small train_lr to finetune.\n* the last fc performs well if it has large output.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmingzhang96%2FMAS-mxnet","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmingzhang96%2FMAS-mxnet","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmingzhang96%2FMAS-mxnet/lists"}