{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:RBIV4AECPIS7DARCVZPOPTJZXN","short_pith_number":"pith:RBIV4AEC","schema_version":"1.0","canonical_sha256":"88515e00827a25f18222ae5ee7cd39bb7d923a926026d076bb6385251160a08f","source":{"kind":"arxiv","id":"2310.05208","version":3},"attestation_state":"computed","paper":{"title":"ZSC-Eval: An Evaluation Toolkit and Benchmark for Multi-agent Zero-shot Coordination","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.HC","cs.LG","cs.MA"],"primary_cat":"cs.AI","authors_text":"Jingxiao Chen, Shao Zhang, Weinan Zhang, Wenhao Zhang, Wentao Dong, Xihuai Wang, Ying Wen","submitted_at":"2023-10-08T15:49:36Z","abstract_excerpt":"Zero-shot coordination (ZSC) is a new cooperative multi-agent reinforcement learning (MARL) challenge that aims to train an ego agent to work with diverse, unseen partners during deployment. The significant difference between the deployment-time partners' distribution and the training partners' distribution determined by the training algorithm makes ZSC a unique out-of-distribution (OOD) generalization challenge. The potential distribution gap between evaluation and deployment-time partners leads to inadequate evaluation, which is exacerbated by the lack of appropriate evaluation metrics. In t"},"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":"2310.05208","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2023-10-08T15:49:36Z","cross_cats_sorted":["cs.HC","cs.LG","cs.MA"],"title_canon_sha256":"6eee79c2715aae925c6f18a6942682eac79fa88a979d3038dc09754431dafb1d","abstract_canon_sha256":"44b908243cc723881a5474f88277035c5f68dbafa7547dc9413a17d7ce4ac0b3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:12:00.874275Z","signature_b64":"vY1XAUryTsCHRO1n+bEwc71GIY0r2X/9pOq+jJYD4LVQiNZXF945mwcueSOwEPeEJWfKzdFtL2DIT4MIpquEAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"88515e00827a25f18222ae5ee7cd39bb7d923a926026d076bb6385251160a08f","last_reissued_at":"2026-07-05T09:12:00.873857Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:12:00.873857Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ZSC-Eval: An Evaluation Toolkit and Benchmark for Multi-agent Zero-shot Coordination","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.HC","cs.LG","cs.MA"],"primary_cat":"cs.AI","authors_text":"Jingxiao Chen, Shao Zhang, Weinan Zhang, Wenhao Zhang, Wentao Dong, Xihuai Wang, Ying Wen","submitted_at":"2023-10-08T15:49:36Z","abstract_excerpt":"Zero-shot coordination (ZSC) is a new cooperative multi-agent reinforcement learning (MARL) challenge that aims to train an ego agent to work with diverse, unseen partners during deployment. The significant difference between the deployment-time partners' distribution and the training partners' distribution determined by the training algorithm makes ZSC a unique out-of-distribution (OOD) generalization challenge. The potential distribution gap between evaluation and deployment-time partners leads to inadequate evaluation, which is exacerbated by the lack of appropriate evaluation metrics. In t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.05208","kind":"arxiv","version":3},"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/2310.05208/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":"2310.05208","created_at":"2026-07-05T09:12:00.873919+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.05208v3","created_at":"2026-07-05T09:12:00.873919+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.05208","created_at":"2026-07-05T09:12:00.873919+00:00"},{"alias_kind":"pith_short_12","alias_value":"RBIV4AECPIS7","created_at":"2026-07-05T09:12:00.873919+00:00"},{"alias_kind":"pith_short_16","alias_value":"RBIV4AECPIS7DARC","created_at":"2026-07-05T09:12:00.873919+00:00"},{"alias_kind":"pith_short_8","alias_value":"RBIV4AEC","created_at":"2026-07-05T09:12:00.873919+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN","json":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN.json","graph_json":"https://pith.science/api/pith-number/RBIV4AECPIS7DARCVZPOPTJZXN/graph.json","events_json":"https://pith.science/api/pith-number/RBIV4AECPIS7DARCVZPOPTJZXN/events.json","paper":"https://pith.science/paper/RBIV4AEC"},"agent_actions":{"view_html":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN","download_json":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN.json","view_paper":"https://pith.science/paper/RBIV4AEC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.05208&json=true","fetch_graph":"https://pith.science/api/pith-number/RBIV4AECPIS7DARCVZPOPTJZXN/graph.json","fetch_events":"https://pith.science/api/pith-number/RBIV4AECPIS7DARCVZPOPTJZXN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN/action/storage_attestation","attest_author":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN/action/author_attestation","sign_citation":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN/action/citation_signature","submit_replication":"https://pith.science/pith/RBIV4AECPIS7DARCVZPOPTJZXN/action/replication_record"}},"created_at":"2026-07-05T09:12:00.873919+00:00","updated_at":"2026-07-05T09:12:00.873919+00:00"}