{"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/quantifying-differences-in-reward-functions","title":"Quantifying Differences in Reward Functions","arxiv_id":"2006.13900","date":"2020-06-24","proceeding":"ICLR 2021 1","authors":["Adam Gleave","Michael Dennis","Shane Legg","Stuart Russell","Jan Leike"],"abstract":"For many tasks, the reward function is inaccessible to introspection or too complex to be specified procedurally, and must instead be learned from user data. Prior work has evaluated learned reward functions by evaluating policies optimized for the learned reward. However, this method cannot distinguish between the learned reward function failing to reflect user preferences and the policy optimization process failing to optimize the learned reward. Moreover, this method can only tell us about behavior in the evaluation environment, but the reward may incentivize very different behavior in even a slightly different deployment environment. To address these problems, we introduce the Equivalent-Policy Invariant Comparison (EPIC) distance to quantify the difference between two reward functions directly, without a policy optimization step. We prove EPIC is invariant on an equivalence class of reward functions that always induce the same optimal policy. Furthermore, we find EPIC can be efficiently approximated and is more robust than baselines to the choice of coverage distribution. Finally, we show that EPIC distance bounds the regret of optimal policies even under different transition dynamics, and we confirm empirically that it predicts policy training success. Our source code is available at https://github.com/HumanCompatibleAI/evaluating-rewards.","url_abs":"https://arxiv.org/abs/2006.13900v3","url_pdf":"https://arxiv.org/pdf/2006.13900v3.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":"quantifying-differences-in-reward-functions","repo_url":"https://github.com/HumanCompatibleAI/evaluating-rewards","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2006.13900","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2006.13900"}},"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/HumanCompatibleAI/evaluating-rewards","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran_draft_wrong":2,"ran_fixture":3,"unverified":10},"by_repo_kind":{"official":{"samples":15,"ran":5,"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":"99673a17240a4b96","entry":"canonical_reward","repo":"humancompatibleai/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/humancompatibleai/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"99673a17240a4b96"}},{"code_sha256_prefix":"57bfe2328f932299","entry":"canonical_reward_distance","repo":"humancompatibleai/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/humancompatibleai/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"57bfe2328f932299"}},{"code_sha256_prefix":"f77296320081f6cb","entry":"canonical_scale_normalizer","repo":"humancompatibleai/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/humancompatibleai/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"f77296320081f6cb"}},{"code_sha256_prefix":"059ae906194fa41e","entry":"direct_distance","repo":"humancompatibleai/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/humancompatibleai/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"059ae906194fa41e"}},{"code_sha256_prefix":"52e8470eebb3f3b2","entry":"lp_norm","repo":"humancompatibleai/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/humancompatibleai/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"52e8470eebb3f3b2"}},{"code_sha256_prefix":"133a3fcd2cec81e8","entry":"_check_dist","repo":"humancompatibleai/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/humancompatibleai/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"133a3fcd2cec81e8"}},{"code_sha256_prefix":"a29d79e85914dfb5","entry":"average_loss","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/analysis/results.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/analysis/results.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a29d79e85914dfb5"}},{"code_sha256_prefix":"0bb7f4c2ffeed496","entry":"average_unwrapped_loss","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/analysis/results.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/analysis/results.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0bb7f4c2ffeed496"}},{"code_sha256_prefix":"de1436200998ee2b","entry":"canonicalize_reward_cfg","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/common_config.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/common_config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"de1436200998ee2b"}},{"code_sha256_prefix":"fde68a593d79622a","entry":"closest_potential","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"fde68a593d79622a"}},{"code_sha256_prefix":"634610fc0258be61","entry":"cross_distance","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/analysis/util.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/analysis/util.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"634610fc0258be61"}},{"code_sha256_prefix":"d51d87c4fe9aac67","entry":"empirical_ci","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/analysis/util.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/analysis/util.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"d51d87c4fe9aac67"}},{"code_sha256_prefix":"305cdb60adc847a4","entry":"point_maze_learned_cfgs","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/common_config.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/common_config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"305cdb60adc847a4"}},{"code_sha256_prefix":"8bf23ee16d057358","entry":"shape","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/distances/tabular.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/distances/tabular.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"8bf23ee16d057358"}},{"code_sha256_prefix":"ae2b77081a249387","entry":"to_series","repo":"HumanCompatibleAI/evaluating-rewards","repo_kind":"official","path":"src/evaluating_rewards/analysis/results.py","file_url":"https://github.com/HumanCompatibleAI/evaluating-rewards/blob/HEAD/src/evaluating_rewards/analysis/results.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ae2b77081a249387"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}