{"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/hiql-offline-goal-conditioned-rl-with-latent-1","title":"HIQL: Offline Goal-Conditioned RL with Latent States as Actions","arxiv_id":"2307.11949","date":"2023-07-22","proceeding":"NeurIPS 2023 11","authors":["Seohong Park","Dibya Ghosh","Benjamin Eysenbach","Sergey Levine"],"abstract":"Unsupervised pre-training has recently become the bedrock for computer vision and natural language processing. In reinforcement learning (RL), goal-conditioned RL can potentially provide an analogous self-supervised approach for making use of large quantities of unlabeled (reward-free) data. However, building effective algorithms for goal-conditioned RL that can learn directly from diverse offline data is challenging, because it is hard to accurately estimate the exact value function for faraway goals. Nonetheless, goal-reaching problems exhibit structure, such that reaching distant goals entails first passing through closer subgoals. This structure can be very useful, as assessing the quality of actions for nearby goals is typically easier than for more distant goals. Based on this idea, we propose a hierarchical algorithm for goal-conditioned RL from offline data. Using one action-free value function, we learn two policies that allow us to exploit this structure: a high-level policy that treats states as actions and predicts (a latent representation of) a subgoal and a low-level policy that predicts the action for reaching this subgoal. Through analysis and didactic examples, we show how this hierarchical decomposition makes our method robust to noise in the estimated value function. We then apply our method to offline goal-reaching benchmarks, showing that our method can solve long-horizon tasks that stymie prior methods, can scale to high-dimensional image observations, and can readily make use of action-free data. Our code is available at https://seohong.me/projects/hiql/","url_abs":"https://arxiv.org/abs/2307.11949v4","url_pdf":"https://arxiv.org/pdf/2307.11949v4.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":"hiql-offline-goal-conditioned-rl-with-latent-1","repo_url":"https://github.com/seohongpark/hiql","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"jax","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"unsupervised-pre-training","task_name":"Unsupervised Pre-training"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2307.11949","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2307.11949"}},"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":"deterministic:regex_extraction","url":"https://github.com/seohongpark/HIQL","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/seohongpark/hiql","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":3,"ran_violates":1,"unverified":13},"by_repo_kind":{"official":{"samples":17,"ran":4,"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":"6ce42ee5c0d1becc","entry":"compute_high_actor_loss","repo":"seohongpark/hiql","repo_kind":"official","path":"src/agents/hiql.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/agents/hiql.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6ce42ee5c0d1becc"}},{"code_sha256_prefix":"33d24c59c871db71","entry":"expectile_loss","repo":"seohongpark/hiql","repo_kind":"official","path":"src/agents/hiql.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/agents/hiql.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"33d24c59c871db71"}},{"code_sha256_prefix":"bfa59d92ee267142","entry":"find_meta","repo":"seohongpark/hiql","repo_kind":"official","path":"calvin/setup_local.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/calvin/setup_local.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"bfa59d92ee267142"}},{"code_sha256_prefix":"5bc947886e7f988c","entry":"is_multi_gpu_training","repo":"seohongpark/HIQL","repo_kind":"official","path":"calvin/calvin_models/calvin_agent/training.py","file_url":"https://github.com/seohongpark/HIQL/blob/HEAD/calvin/calvin_models/calvin_agent/training.py","link_basis":"first_harvest_node","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5bc947886e7f988c"}},{"code_sha256_prefix":"77ee8e1bcb496897","entry":"compute_actor_loss","repo":"seohongpark/hiql","repo_kind":"official","path":"src/agents/hiql.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/agents/hiql.py","link_basis":"first_harvest_node","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":"77ee8e1bcb496897"}},{"code_sha256_prefix":"7f174e0d2fe49c08","entry":"ensemblize","repo":"seohongpark/hiql","repo_kind":"official","path":"jaxrl_m/networks.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/jaxrl_m/networks.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":"7f174e0d2fe49c08"}},{"code_sha256_prefix":"2d17f9689108389d","entry":"get_canvas_image","repo":"seohongpark/hiql","repo_kind":"official","path":"src/d4rl_ant.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/d4rl_ant.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":"2d17f9689108389d"}},{"code_sha256_prefix":"e60bfdd0574ddc5f","entry":"get_normalization","repo":"seohongpark/hiql","repo_kind":"official","path":"src/d4rl_utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/d4rl_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":true,"mcp_get_code":{"code_sha256":"e60bfdd0574ddc5f"}},{"code_sha256_prefix":"aa92173113012077","entry":"interp2d","repo":"seohongpark/hiql","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/utils.py","link_basis":"first_harvest_node","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":"aa92173113012077"}},{"code_sha256_prefix":"9e273495210cbc4c","entry":"make_visual","repo":"seohongpark/hiql","repo_kind":"official","path":"src/viz_utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/viz_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":true,"mcp_get_code":{"code_sha256":"9e273495210cbc4c"}},{"code_sha256_prefix":"8bc4244b8915e237","entry":"make_visual_no_image","repo":"seohongpark/hiql","repo_kind":"official","path":"src/viz_utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/viz_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":true,"mcp_get_code":{"code_sha256":"8bc4244b8915e237"}},{"code_sha256_prefix":"fd2c92d2074f0195","entry":"most_squarelike","repo":"seohongpark/hiql","repo_kind":"official","path":"src/viz_utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/viz_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":true,"mcp_get_code":{"code_sha256":"fd2c92d2074f0195"}},{"code_sha256_prefix":"593372873bfa1f9c","entry":"prepare_video","repo":"seohongpark/hiql","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/utils.py","link_basis":"first_harvest_node","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":"593372873bfa1f9c"}},{"code_sha256_prefix":"2d208c5ae7bb63f6","entry":"save_video","repo":"seohongpark/hiql","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/utils.py","link_basis":"first_harvest_node","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":"2d208c5ae7bb63f6"}},{"code_sha256_prefix":"a5514c312d961e08","entry":"valid_goal_sampler","repo":"seohongpark/hiql","repo_kind":"official","path":"src/d4rl_ant.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/d4rl_ant.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":"a5514c312d961e08"}},{"code_sha256_prefix":"7bc5beaaf5ef3e71","entry":"visualize_gcvalues","repo":"seohongpark/hiql","repo_kind":"official","path":"src/roomworld_utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/roomworld_utils.py","link_basis":"first_harvest_node","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":"7bc5beaaf5ef3e71"}},{"code_sha256_prefix":"220888da8c3ceab5","entry":"visualize_trajectories","repo":"seohongpark/hiql","repo_kind":"official","path":"src/roomworld_utils.py","file_url":"https://github.com/seohongpark/hiql/blob/HEAD/src/roomworld_utils.py","link_basis":"first_harvest_node","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":"220888da8c3ceab5"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}