{"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/overcoming-exploration-in-reinforcement","title":"Overcoming Exploration in Reinforcement Learning with Demonstrations","arxiv_id":"1709.10089","date":"2017-09-28","proceeding":null,"authors":["Ashvin Nair","Bob McGrew","Marcin Andrychowicz","Wojciech Zaremba","Pieter Abbeel"],"abstract":"Exploration in environments with sparse rewards has been a persistent problem\nin reinforcement learning (RL). Many tasks are natural to specify with a sparse\nreward, and manually shaping a reward function can result in suboptimal\nperformance. However, finding a non-zero reward is exponentially more difficult\nwith increasing task horizon or action dimensionality. This puts many\nreal-world tasks out of practical reach of RL methods. In this work, we use\ndemonstrations to overcome the exploration problem and successfully learn to\nperform long-horizon, multi-step robotics tasks with continuous control such as\nstacking blocks with a robot arm. Our method, which builds on top of Deep\nDeterministic Policy Gradients and Hindsight Experience Replay, provides an\norder of magnitude of speedup over RL on simulated robotics tasks. It is simple\nto implement and makes only the additional assumption that we can collect a\nsmall set of demonstrations. Furthermore, our method is able to solve tasks not\nsolvable by either RL or behavior cloning alone, and often ends up\noutperforming the demonstrator policy.","url_abs":"http://arxiv.org/abs/1709.10089v2","url_pdf":"http://arxiv.org/pdf/1709.10089v2.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":"overcoming-exploration-in-reinforcement","repo_url":"https://github.com/MrSyee/pg-is-all-you-need","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"overcoming-exploration-in-reinforcement","repo_url":"https://github.com/jangirrishabh/Overcoming-exploration-from-demos","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"overcoming-exploration-in-reinforcement","repo_url":"https://github.com/kaixindelele/DRLib","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"continuous-control","task_name":"Continuous Control"},{"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"}],"methods":[{"method_slug":"experience-replay","method_name":"Experience Replay"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1709.10089","atlas_url":"https://app.syntology.ai/?focus=1709.10089","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1709.10089"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+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/MrSyee/pg-is-all-you-need","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jangirrishabh/Overcoming-exploration-from-demos","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kaixindelele/DRLib","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran_draft_wrong":2,"unverified":3},"by_repo_kind":{"listed":{"samples":4,"ran":1,"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":1,"samples":[{"code_sha256_prefix":"a68f6550efbfd9a5","entry":"dims_to_shapes","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"a68f6550efbfd9a5"}},{"code_sha256_prefix":"e0785843af667cdd","entry":"make_sample_her_transitions","repo":"jangirrishabh/Overcoming-exploration-from-demos","repo_kind":"listed","path":"her.py","file_url":"https://github.com/jangirrishabh/Overcoming-exploration-from-demos/blob/HEAD/her.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e0785843af667cdd"}},{"code_sha256_prefix":"6874fb8652a4c8a3","entry":"add_noise_obs2state","repo":"kaixindelele/DRLib","repo_kind":"listed","path":"D2SSR/d2ssr_train_torch.py","file_url":"https://github.com/kaixindelele/DRLib/blob/HEAD/D2SSR/d2ssr_train_torch.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":"6874fb8652a4c8a3"}},{"code_sha256_prefix":"6a2c71b97236af90","entry":"obs2state","repo":"kaixindelele/DRLib","repo_kind":"listed","path":"train_tf1.py","file_url":"https://github.com/kaixindelele/DRLib/blob/HEAD/train_tf1.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":"6a2c71b97236af90"}},{"code_sha256_prefix":"bd60c9cb106e111f","entry":"obs2state","repo":"kaixindelele/DRLib","repo_kind":"listed","path":"D2SSR/d2ssr_train_torch.py","file_url":"https://github.com/kaixindelele/DRLib/blob/HEAD/D2SSR/d2ssr_train_torch.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":"bd60c9cb106e111f"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}