{"id":15906722,"url":"https://github.com/csinva/max-activation-interpretation-pytorch","last_synced_at":"2025-08-14T18:32:08.479Z","repository":{"id":97069810,"uuid":"179602489","full_name":"csinva/max-activation-interpretation-pytorch","owner":"csinva","description":"Code for creating maximal activation images (like Deep Dream) in pytorch with various regularizations / losses.","archived":false,"fork":false,"pushed_at":"2020-01-31T00:00:15.000Z","size":193,"stargazers_count":4,"open_issues_count":0,"forks_count":2,"subscribers_count":3,"default_branch":"master","last_synced_at":"2024-10-28T11:35:37.674Z","etag":null,"topics":["deep-dream","deep-learning","interpretability","maximal-activation","neural-network","optimization","pytorch","regularization","total-variation","visualization-tools"],"latest_commit_sha":null,"homepage":null,"language":"Jupyter Notebook","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/csinva.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}},"created_at":"2019-04-05T01:31:26.000Z","updated_at":"2024-03-04T08:39:13.000Z","dependencies_parsed_at":"2023-03-08T11:15:26.314Z","dependency_job_id":null,"html_url":"https://github.com/csinva/max-activation-interpretation-pytorch","commit_stats":{"total_commits":10,"total_committers":1,"mean_commits":10.0,"dds":0.0,"last_synced_commit":"bafc6a69c05be3a60bc6473953a91ccf4173fa8f"},"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/csinva%2Fmax-activation-interpretation-pytorch","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/csinva%2Fmax-activation-interpretation-pytorch/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/csinva%2Fmax-activation-interpretation-pytorch/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/csinva%2Fmax-activation-interpretation-pytorch/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/csinva","download_url":"https://codeload.github.com/csinva/max-activation-interpretation-pytorch/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":229854758,"owners_count":18134829,"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":["deep-dream","deep-learning","interpretability","maximal-activation","neural-network","optimization","pytorch","regularization","total-variation","visualization-tools"],"created_at":"2024-10-06T13:41:31.008Z","updated_at":"2024-12-15T17:41:45.195Z","avatar_url":"https://github.com/csinva.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# maximal activation\n- maximal activation is a simple technique which optimizes the input of a model to maximize an output response\n- the code here (in `max_act.py`) shows a simple pytorch implementation of this technique\n- this code includes simple regularization for this method\n- one example maximizing the class \"peacock\" for AlexNet: \n\n![](ims/peacock.png)\n\n\n## sample usage\n\n- install with `pip install git+https://github.com/csinva/max-activation-interpretation-pytorch`\n\n```python\nsys.path.append('../max_act')\nfrom max_act import maximize_im, maximize_im_simple\nimport visualize_ims as viz\n\ndevice = 'cuda'\nmodel = model.to(device)\nclass_num = 5\nim_shape = (1, 1, 28, 28) # (1, 3, 224, 224) for imagenet\nim = torch.zeros(im_shape, requires_grad=True, device=device)\nims_opt, losses = maximize_im_simple(model, im, class_num=class_num, lr=1e-5,\n                                     num_iters=int(1e3), lambda_tv=1e-1, lambda_pnorm=1e-1)\n\nviz.show(ims_opt[::2])\nplt.show()\n\nplt.plot(losses)\n```","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcsinva%2Fmax-activation-interpretation-pytorch","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fcsinva%2Fmax-activation-interpretation-pytorch","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fcsinva%2Fmax-activation-interpretation-pytorch/lists"}