{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/generating-neural-networks-with-neural","title":"Generating Neural Networks with Neural Networks","arxiv_id":"1801.01952","date":"2018-01-06","proceeding":null,"authors":["Lior Deutsch"],"abstract":"Hypernetworks are neural networks that generate weights for another neural\nnetwork. We formulate the hypernetwork training objective as a compromise\nbetween accuracy and diversity, where the diversity takes into account trivial\nsymmetry transformations of the target network. We explain how this simple\nformulation generalizes variational inference. We use multi-layered perceptrons\nto form the mapping from the low dimensional input random vector to the high\ndimensional weight space, and demonstrate how to reduce the number of\nparameters in this mapping by parameter sharing. We perform experiments and\nshow that the generated weights are diverse and lie on a non-trivial manifold.","url_abs":"http://arxiv.org/abs/1801.01952v4","url_pdf":"http://arxiv.org/pdf/1801.01952v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"generating-neural-networks-with-neural","repo_url":"https://github.com/sliorde/generating-neural-networks-with-neural-networks","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[{"method_slug":"hypernetwork","method_name":"HyperNetwork"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1801.01952","atlas_url":"https://app.syntology.ai/?focus=1801.01952","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}