{"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/backpropagation-through-the-void-optimizing","title":"Backpropagation through the Void: Optimizing control variates for black-box gradient estimation","arxiv_id":"1711.00123","date":"2017-10-31","proceeding":"ICLR 2018 1","authors":["Will Grathwohl","Dami Choi","Yuhuai Wu","Geoffrey Roeder","David Duvenaud"],"abstract":"Gradient-based optimization is the foundation of deep learning and\nreinforcement learning. Even when the mechanism being optimized is unknown or\nnot differentiable, optimization using high-variance or biased gradient\nestimates is still often the best strategy. We introduce a general framework\nfor learning low-variance, unbiased gradient estimators for black-box functions\nof random variables. Our method uses gradients of a neural network trained\njointly with model parameters or policies, and is applicable in both discrete\nand continuous settings. We demonstrate this framework for training discrete\nlatent-variable models. We also give an unbiased, action-conditional extension\nof the advantage actor-critic reinforcement learning algorithm.","url_abs":"http://arxiv.org/abs/1711.00123v3","url_pdf":"http://arxiv.org/pdf/1711.00123v3.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":"backpropagation-through-the-void-optimizing","repo_url":"https://github.com/duvenaud/relax","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"backpropagation-through-the-void-optimizing","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":"backpropagation-through-the-void-optimizing","repo_url":"https://github.com/brain-research/mirage-rl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"backpropagation-through-the-void-optimizing","repo_url":"https://github.com/thlautenschlaeger/bpttv-lax","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"backpropagation-through-the-void-optimizing","repo_url":"https://github.com/Bonnevie/rebar","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"backpropagation-through-the-void-optimizing","repo_url":"https://github.com/ElleryL/gradient_estimator","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null},{"paper_slug":"backpropagation-through-the-void-optimizing","repo_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.00123","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1711.00123"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/thlautenschlaeger/bpttv-lax","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/brain-research/mirage-rl","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/duvenaud/relax","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ElleryL/gradient_estimator","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Bonnevie/rebar","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/TalkToTheGAN/REGAN","reach":{"status":"ok"}}],"summary":{"ran_honours":3,"ran_violates":1,"unverified":8},"by_repo_kind":{"official":{"samples":1,"ran":1,"repositories":1},"listed":{"samples":11,"ran":3,"repositories":2}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":12,"samples":[{"code_sha256_prefix":"b8a97b9a9a17b91a","entry":"Adam_Optim","repo":"ElleryL/gradient_estimator","repo_kind":"listed","path":"my_reLAX.py","file_url":"https://github.com/ElleryL/gradient_estimator/blob/HEAD/my_reLAX.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b8a97b9a9a17b91a"}},{"code_sha256_prefix":"4b29bd089ba1c204","entry":"d_log_q_z_given_x","repo":"ElleryL/gradient_estimator","repo_kind":"listed","path":"my_reLAX.py","file_url":"https://github.com/ElleryL/gradient_estimator/blob/HEAD/my_reLAX.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"4b29bd089ba1c204"}},{"code_sha256_prefix":"3e154795b809f8c9","entry":"heaviside","repo":"duvenaud/relax","repo_kind":"official","path":"relax-autograd/relax.py","file_url":"https://github.com/duvenaud/relax/blob/HEAD/relax-autograd/relax.py","link_basis":"first_harvest_node","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3e154795b809f8c9"}},{"code_sha256_prefix":"e54e926116f5ce7e","entry":"prob_mapping","repo":"ElleryL/gradient_estimator","repo_kind":"listed","path":"my_reLAX.py","file_url":"https://github.com/ElleryL/gradient_estimator/blob/HEAD/my_reLAX.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"e54e926116f5ce7e"}},{"code_sha256_prefix":"dafc66df201789e0","entry":"cat_entropy","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/a2c/utils.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/a2c/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"dafc66df201789e0"}},{"code_sha256_prefix":"7082b4bbc0d9384f","entry":"cat_entropy_softmax","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/a2c/utils.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/a2c/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"7082b4bbc0d9384f"}},{"code_sha256_prefix":"927e1047481e9318","entry":"gs","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/a2c/a2c.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/a2c/a2c.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"927e1047481e9318"}},{"code_sha256_prefix":"cd68a1f53db269d4","entry":"make_samples","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/a2c/policies.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/a2c/policies.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"cd68a1f53db269d4"}},{"code_sha256_prefix":"aad51e2df5a9fb5d","entry":"pathlength","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/a2c/a2c_cont.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/a2c/a2c_cont.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"aad51e2df5a9fb5d"}},{"code_sha256_prefix":"68b23feea13c46c8","entry":"read_csv","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/logger.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/logger.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"68b23feea13c46c8"}},{"code_sha256_prefix":"88ef5a17fc986d9a","entry":"read_json","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/logger.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/logger.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"88ef5a17fc986d9a"}},{"code_sha256_prefix":"178099b1f8fb2b9d","entry":"sample","repo":"wgrathwohl/BackpropThroughTheVoidRL","repo_kind":"listed","path":"baselines/a2c/utils.py","file_url":"https://github.com/wgrathwohl/BackpropThroughTheVoidRL/blob/HEAD/baselines/a2c/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"178099b1f8fb2b9d"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}