{"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/diagnosing-bottlenecks-in-deep-q-learning","title":"Diagnosing Bottlenecks in Deep Q-learning Algorithms","arxiv_id":"1902.10250","date":"2019-02-26","proceeding":null,"authors":["Justin Fu","Aviral Kumar","Matthew Soh","Sergey Levine"],"abstract":"Q-learning methods represent a commonly used class of algorithms in\nreinforcement learning: they are generally efficient and simple, and can be\ncombined readily with function approximators for deep reinforcement learning\n(RL). However, the behavior of Q-learning methods with function approximation\nis poorly understood, both theoretically and empirically. In this work, we aim\nto experimentally investigate potential issues in Q-learning, by means of a\n\"unit testing\" framework where we can utilize oracles to disentangle sources of\nerror. Specifically, we investigate questions related to function\napproximation, sampling error and nonstationarity, and where available, verify\nif trends found in oracle settings hold true with modern deep RL methods. We\nfind that large neural network architectures have many benefits with regards to\nlearning stability; offer several practical compensations for overfitting; and\ndevelop a novel sampling method based on explicitly compensating for function\napproximation error that yields fair improvement on high-dimensional continuous\ncontrol domains.","url_abs":"http://arxiv.org/abs/1902.10250v1","url_pdf":"http://arxiv.org/pdf/1902.10250v1.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":"diagnosing-bottlenecks-in-deep-q-learning","repo_url":"https://github.com/justinjfu/diagnosing_qlearning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"continuous-control","task_name":"Continuous Control"},{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"q-learning","task_name":"Q-Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"continuous-control","task_name":"continuous-control"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[{"method_slug":"q-learning","method_name":"Q-Learning"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.10250","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1902.10250"}},"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/justinjfu/diagnosing_qlearning","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"unverified":10},"by_repo_kind":{"listed":{"samples":10,"ran":0,"repositories":1}},"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":0,"samples":[{"code_sha256_prefix":"f772ab853b7d7946","entry":"DictArray","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"rlutil/dictarray.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/rlutil/dictarray.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":true,"mcp_get_code":{"code_sha256":"f772ab853b7d7946"}},{"code_sha256_prefix":"e42fbeb98e676352","entry":"all_tensor","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"debugq/pytorch_util.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/debugq/pytorch_util.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":true,"mcp_get_code":{"code_sha256":"e42fbeb98e676352"}},{"code_sha256_prefix":"aa1b3956fafbf750","entry":"always_true","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"rlutil/hyper_sweep.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/rlutil/hyper_sweep.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":true,"mcp_get_code":{"code_sha256":"aa1b3956fafbf750"}},{"code_sha256_prefix":"241dcd3902fa8170","entry":"chunk_filter","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"rlutil/hyper_sweep.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/rlutil/hyper_sweep.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":true,"mcp_get_code":{"code_sha256":"241dcd3902fa8170"}},{"code_sha256_prefix":"60d4b81474c1cd33","entry":"flatten_list","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"rlutil/general.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/rlutil/general.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":true,"mcp_get_code":{"code_sha256":"60d4b81474c1cd33"}},{"code_sha256_prefix":"23202e963c5fa970","entry":"kwargs_wrapper","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"rlutil/hyper_sweep.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/rlutil/hyper_sweep.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":true,"mcp_get_code":{"code_sha256":"23202e963c5fa970"}},{"code_sha256_prefix":"67abccd8c00e909e","entry":"stack_observations","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"debugq/algos/sampling_fqi.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/debugq/algos/sampling_fqi.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":true,"mcp_get_code":{"code_sha256":"67abccd8c00e909e"}},{"code_sha256_prefix":"8f062b4b0c962b27","entry":"tensor","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"debugq/pytorch_util.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/debugq/pytorch_util.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":true,"mcp_get_code":{"code_sha256":"8f062b4b0c962b27"}},{"code_sha256_prefix":"2d42b4d3a6515d97","entry":"to_numpy","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"debugq/pytorch_util.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/debugq/pytorch_util.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":true,"mcp_get_code":{"code_sha256":"2d42b4d3a6515d97"}},{"code_sha256_prefix":"cae60f0f73bb7947","entry":"weighted_q_diff","repo":"justinjfu/diagnosing_qlearning","repo_kind":"listed","path":"debugq/algos/fqi.py","file_url":"https://github.com/justinjfu/diagnosing_qlearning/blob/HEAD/debugq/algos/fqi.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":true,"mcp_get_code":{"code_sha256":"cae60f0f73bb7947"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}