{"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/rebar-low-variance-unbiased-gradient","title":"REBAR: Low-variance, unbiased gradient estimates for discrete latent variable models","arxiv_id":"1703.07370","date":"2017-03-21","proceeding":"NeurIPS 2017 12","authors":["George Tucker","andriy mnih","Chris J. Maddison","Dieterich Lawson","Jascha Sohl-Dickstein"],"abstract":"Learning in models with discrete latent variables is challenging due to high\nvariance gradient estimators. Generally, approaches have relied on control\nvariates to reduce the variance of the REINFORCE estimator. Recent work (Jang\net al. 2016, Maddison et al. 2016) has taken a different approach, introducing\na continuous relaxation of discrete variables to produce low-variance, but\nbiased, gradient estimates. In this work, we combine the two approaches through\na novel control variate that produces low-variance, \\emph{unbiased} gradient\nestimates. Then, we introduce a modification to the continuous relaxation and\nshow that the tightness of the relaxation can be adapted online, removing it as\na hyperparameter. We show state-of-the-art variance reduction on several\nbenchmark generative modeling tasks, generally leading to faster convergence to\na better final log-likelihood.","url_abs":"http://arxiv.org/abs/1703.07370v4","url_pdf":"http://arxiv.org/pdf/1703.07370v4.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":"rebar-low-variance-unbiased-gradient","repo_url":"https://github.com/tensorflow/models","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null},{"paper_slug":"rebar-low-variance-unbiased-gradient","repo_url":"https://github.com/TalkToTheGAN/REGAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"rebar-low-variance-unbiased-gradient","repo_url":"https://github.com/tensorflow/models/tree/master/research/rebar","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[],"methods":[{"method_slug":"reinforce","method_name":"REINFORCE"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1703.07370","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}