{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UITKPTYGBZQGBV4TLRT7XW6FWS","short_pith_number":"pith:UITKPTYG","schema_version":"1.0","canonical_sha256":"a226a7cf060e6060d7935c67fbdbc5b48fb8343f04554ba5fbf31add9353a963","source":{"kind":"arxiv","id":"2504.15785","version":1},"attestation_state":"computed","paper":{"title":"WALL-E 2.0: World Alignment by NeuroSymbolic Learning improves World Model-based LLM Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chengqi Zhang, Deheng Ye, Guodong Long, Jing Jiang, Siyu Zhou, Tianyi Zhou, Yijun Yang","submitted_at":"2025-04-22T10:58:27Z","abstract_excerpt":"Can we build accurate world models out of large language models (LLMs)? How can world models benefit LLM agents? The gap between the prior knowledge of LLMs and the specified environment's dynamics usually bottlenecks LLMs' performance as world models. To bridge the gap, we propose a training-free \"world alignment\" that learns an environment's symbolic knowledge complementary to LLMs. The symbolic knowledge covers action rules, knowledge graphs, and scene graphs, which are extracted by LLMs from exploration trajectories and encoded into executable codes to regulate LLM agents' policies. We fur"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2504.15785","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-04-22T10:58:27Z","cross_cats_sorted":[],"title_canon_sha256":"d7fc4b853323ab21a51350d117b83e1699836c7c569f4b7883a6e935fbd64ed4","abstract_canon_sha256":"9b919ade85440caae04bb37b1c8e3a66ba087d698aec5197fa7f973cd5107238"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:52:29.946655Z","signature_b64":"fcK8ka7AdJ+leBUjG0vXFx7FvL5kUoICYBlb1ZC5JQYizE+bO8KJJlBFKfwMh++0OZIOixcKHiaRvObV1CefBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a226a7cf060e6060d7935c67fbdbc5b48fb8343f04554ba5fbf31add9353a963","last_reissued_at":"2026-07-05T10:52:29.946143Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:52:29.946143Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"WALL-E 2.0: World Alignment by NeuroSymbolic Learning improves World Model-based LLM Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chengqi Zhang, Deheng Ye, Guodong Long, Jing Jiang, Siyu Zhou, Tianyi Zhou, Yijun Yang","submitted_at":"2025-04-22T10:58:27Z","abstract_excerpt":"Can we build accurate world models out of large language models (LLMs)? How can world models benefit LLM agents? The gap between the prior knowledge of LLMs and the specified environment's dynamics usually bottlenecks LLMs' performance as world models. To bridge the gap, we propose a training-free \"world alignment\" that learns an environment's symbolic knowledge complementary to LLMs. The symbolic knowledge covers action rules, knowledge graphs, and scene graphs, which are extracted by LLMs from exploration trajectories and encoded into executable codes to regulate LLM agents' policies. We fur"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15785","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2504.15785/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2504.15785","created_at":"2026-07-05T10:52:29.946203+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15785v1","created_at":"2026-07-05T10:52:29.946203+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15785","created_at":"2026-07-05T10:52:29.946203+00:00"},{"alias_kind":"pith_short_12","alias_value":"UITKPTYGBZQG","created_at":"2026-07-05T10:52:29.946203+00:00"},{"alias_kind":"pith_short_16","alias_value":"UITKPTYGBZQGBV4T","created_at":"2026-07-05T10:52:29.946203+00:00"},{"alias_kind":"pith_short_8","alias_value":"UITKPTYG","created_at":"2026-07-05T10:52:29.946203+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.16725","citing_title":"Baba in Wonderland: Online Self-Supervised Dynamics Discovery for Executable World Models","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14175","citing_title":"Grounded Continuation: A Linear-Time Runtime Verifier for LLM Conversations","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09487","citing_title":"Kintsugi: Learning Policies by Repairing Executable Knowledge Bases","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS","json":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS.json","graph_json":"https://pith.science/api/pith-number/UITKPTYGBZQGBV4TLRT7XW6FWS/graph.json","events_json":"https://pith.science/api/pith-number/UITKPTYGBZQGBV4TLRT7XW6FWS/events.json","paper":"https://pith.science/paper/UITKPTYG"},"agent_actions":{"view_html":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS","download_json":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS.json","view_paper":"https://pith.science/paper/UITKPTYG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15785&json=true","fetch_graph":"https://pith.science/api/pith-number/UITKPTYGBZQGBV4TLRT7XW6FWS/graph.json","fetch_events":"https://pith.science/api/pith-number/UITKPTYGBZQGBV4TLRT7XW6FWS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS/action/storage_attestation","attest_author":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS/action/author_attestation","sign_citation":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS/action/citation_signature","submit_replication":"https://pith.science/pith/UITKPTYGBZQGBV4TLRT7XW6FWS/action/replication_record"}},"created_at":"2026-07-05T10:52:29.946203+00:00","updated_at":"2026-07-05T10:52:29.946203+00:00"}