{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:4RANQRIG7CIP6ZY4O2KYPT54NN","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"ad3784548ded4a14b5fafc6dbc128558fb41c2cdec7eae24949d70c5961dbb6c","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2023-10-19T17:31:01Z","title_canon_sha256":"cb5b85979677dc841ed11123093ae61df2a02fa38dcba4fe26ffe246c855c92e"},"schema_version":"1.0","source":{"id":"2310.12931","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.12931","created_at":"2026-05-17T23:39:21Z"},{"alias_kind":"arxiv_version","alias_value":"2310.12931v2","created_at":"2026-05-17T23:39:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.12931","created_at":"2026-05-17T23:39:21Z"},{"alias_kind":"pith_short_12","alias_value":"4RANQRIG7CIP","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"4RANQRIG7CIP6ZY4","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"4RANQRIG","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:3eb63a242762c6828f9886a244e764f4afd66324553a82096fe93035cfa18eab","target":"graph","created_at":"2026-05-17T23:39:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Without any task-specific prompting or pre-defined reward templates, Eureka generates reward functions that outperform expert human-engineered rewards. In a diverse suite of 29 open-source RL environments that include 10 distinct robot morphologies, Eureka outperforms human experts on 83% of the tasks, leading to an average normalized improvement of 52%."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That LLM-generated reward code will continue to produce stable, non-degenerate policies when transferred to new tasks or real hardware, rather than exploiting simulator quirks that do not appear in the reported 29 environments."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Eureka uses LLMs for evolutionary optimization of reward code to outperform human experts on 83% of 29 RL tasks with 52% average improvement and enables gradient-free RLHF."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Large language models can design reward functions for robot tasks that outperform those created by human experts."}],"snapshot_sha256":"400320205c025bb9e78d1d801867ebf301ca5aefe735e9adec57308f006a3542"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f69cc1d530b44f7709403e4ba609fffa2a572d54bca81a04ba0d6f8219c438f8"},"paper":{"abstract_excerpt":"Large Language Models (LLMs) have excelled as high-level semantic planners for sequential decision-making tasks. However, harnessing them to learn complex low-level manipulation tasks, such as dexterous pen spinning, remains an open problem. We bridge this fundamental gap and present Eureka, a human-level reward design algorithm powered by LLMs. Eureka exploits the remarkable zero-shot generation, code-writing, and in-context improvement capabilities of state-of-the-art LLMs, such as GPT-4, to perform evolutionary optimization over reward code. The resulting rewards can then be used to acquire","authors_text":"Anima Anandkumar, De-An Huang, Dinesh Jayaraman, Guanzhi Wang, Linxi Fan, Osbert Bastani, William Liang, Yecheng Jason Ma, Yuke Zhu","cross_cats":["cs.AI","cs.LG"],"headline":"Large language models can design reward functions for robot tasks that outperform those created by human experts.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2023-10-19T17:31:01Z","title":"Eureka: Human-Level Reward Design via Coding Large Language Models"},"references":{"count":14,"internal_anchors":0,"resolved_work":14,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"If you see phrases like [NUM: default_value], replace the entire phrase with a numerical value","work_id":"e1404de5-7f99-4b50-b730-b8dd755a9245","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"If you see phrases like {CHOICE: choice1, choice2, ...}, it means you should replace the entire phrase with one of the choices listed","work_id":"08d69275-c02c-44ab-ac69-f6ba9199ff18","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"If you see [optional], it means you only add that line if necessary for the task, otherwise remove that line","work_id":"cd6a8140-9399-4f50-b78d-ffda67ed8f68","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Do not invent new objects not listed here","work_id":"9aa3520b-1406-4dc3-b23d-d4453816fbc5","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Always start the description with [start of plan] and end it with [end of plan]","work_id":"76f847c9-fc23-4dda-927a-262b3d2c5909","year":null}],"snapshot_sha256":"e0f3eb979b31a037bc6b19dae15be2ca86d18273a46c2bba621f617a7a72af06"},"source":{"id":"2310.12931","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-14T20:10:45.963363Z","id":"521bdc28-8cd4-4e95-be37-5921a4011a8c","model_set":{"reader":"grok-4.3"},"one_line_summary":"Eureka uses LLMs for evolutionary optimization of reward code to outperform human experts on 83% of 29 RL tasks with 52% average improvement and enables gradient-free RLHF.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Large language models can design reward functions for robot tasks that outperform those created by human experts.","strongest_claim":"Without any task-specific prompting or pre-defined reward templates, Eureka generates reward functions that outperform expert human-engineered rewards. In a diverse suite of 29 open-source RL environments that include 10 distinct robot morphologies, Eureka outperforms human experts on 83% of the tasks, leading to an average normalized improvement of 52%.","weakest_assumption":"That LLM-generated reward code will continue to produce stable, non-degenerate policies when transferred to new tasks or real hardware, rather than exploiting simulator quirks that do not appear in the reported 29 environments."}},"verdict_id":"521bdc28-8cd4-4e95-be37-5921a4011a8c"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4a8e303869c353064d500d3c4262497b40da26c18ad668af9702f83c8dd41ad0","target":"record","created_at":"2026-05-17T23:39:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"ad3784548ded4a14b5fafc6dbc128558fb41c2cdec7eae24949d70c5961dbb6c","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2023-10-19T17:31:01Z","title_canon_sha256":"cb5b85979677dc841ed11123093ae61df2a02fa38dcba4fe26ffe246c855c92e"},"schema_version":"1.0","source":{"id":"2310.12931","kind":"arxiv","version":2}},"canonical_sha256":"e440d84506f890ff671c769587cfbc6b63cefa2aa6492989b34337cb0c3515af","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e440d84506f890ff671c769587cfbc6b63cefa2aa6492989b34337cb0c3515af","first_computed_at":"2026-05-17T23:39:21.855779Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:21.855779Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"1Xq3VKjvDfXAs4cf0HHgQvEoKAAmg6zno7I5qsNurEweK39ebaTpct3exCEcM/LP2ICtNAb9FbCSJie3dUImDQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:21.856546Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.12931","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4a8e303869c353064d500d3c4262497b40da26c18ad668af9702f83c8dd41ad0","sha256:3eb63a242762c6828f9886a244e764f4afd66324553a82096fe93035cfa18eab"],"state_sha256":"772f3fce83c4301cfb35d292172b340b0fd11ae15b8958eaaa9ddcbc40e1f0ef"}