{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:G6C253PIK2ZU3TUZ2FPCTKW26P","short_pith_number":"pith:G6C253PI","schema_version":"1.0","canonical_sha256":"3785aeede856b34dce99d15e29aadaf3ef774fae8c94975e5cf30ff78380719f","source":{"kind":"arxiv","id":"2406.18872","version":2},"attestation_state":"computed","paper":{"title":"Efficacy of Language Model Self-Play in Non-Zero-Sum Games","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Austen Liao, Dan Klein, Nicholas Tomlin","submitted_at":"2024-06-27T03:52:35Z","abstract_excerpt":"Game-playing agents like AlphaGo have achieved superhuman performance through self-play, which is theoretically guaranteed to yield optimal policies in competitive games. However, most language tasks are partially or fully cooperative, so it is an open question whether techniques like self-play can effectively be used to improve language models. We empirically investigate this question in a negotiation game setting known as Deal or No Deal (DoND). Crucially, the objective in DoND can be modified to produce a fully cooperative game, a strictly competitive one, or anything in between. We finetun"},"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":"2406.18872","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-27T03:52:35Z","cross_cats_sorted":[],"title_canon_sha256":"98329a18db8e31306d4fc6ceb5f8fe00680d4aea3173c685f87e8ff4ac380174","abstract_canon_sha256":"fbddf761c66040086436c0ac44e26671e763e2d32e4f7ccf849bedc06e2b949e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:45:50.813257Z","signature_b64":"0pc0A3rSODf2TeWir6TCX512lLQg3eoeq27XR+iJyd+esQ66XshRIBt9eKjqgT35rBZu3fx8Q9RPhIAD5uI9AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3785aeede856b34dce99d15e29aadaf3ef774fae8c94975e5cf30ff78380719f","last_reissued_at":"2026-07-05T09:45:50.812803Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:45:50.812803Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficacy of Language Model Self-Play in Non-Zero-Sum Games","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Austen Liao, Dan Klein, Nicholas Tomlin","submitted_at":"2024-06-27T03:52:35Z","abstract_excerpt":"Game-playing agents like AlphaGo have achieved superhuman performance through self-play, which is theoretically guaranteed to yield optimal policies in competitive games. However, most language tasks are partially or fully cooperative, so it is an open question whether techniques like self-play can effectively be used to improve language models. We empirically investigate this question in a negotiation game setting known as Deal or No Deal (DoND). Crucially, the objective in DoND can be modified to produce a fully cooperative game, a strictly competitive one, or anything in between. We finetun"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.18872","kind":"arxiv","version":2},"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/2406.18872/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":"2406.18872","created_at":"2026-07-05T09:45:50.812860+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.18872v2","created_at":"2026-07-05T09:45:50.812860+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.18872","created_at":"2026-07-05T09:45:50.812860+00:00"},{"alias_kind":"pith_short_12","alias_value":"G6C253PIK2ZU","created_at":"2026-07-05T09:45:50.812860+00:00"},{"alias_kind":"pith_short_16","alias_value":"G6C253PIK2ZU3TUZ","created_at":"2026-07-05T09:45:50.812860+00:00"},{"alias_kind":"pith_short_8","alias_value":"G6C253PI","created_at":"2026-07-05T09:45:50.812860+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17657","citing_title":"Using Cognitive Models to Improve Language Model Simulation of Human Persuasion Games","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01750","citing_title":"Talk is Cheap, Communication is Hard: Dynamic Grounding Failures and Repair in Multi-Agent Negotiation","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08427","citing_title":"The Attacker in the Mirror: Breaking Self-Consistency in Safety via Anchored Bipolicy Self-Play","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08327","citing_title":"Interactive Critique-Revision Training for Reliable Structured LLM Generation","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P","json":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P.json","graph_json":"https://pith.science/api/pith-number/G6C253PIK2ZU3TUZ2FPCTKW26P/graph.json","events_json":"https://pith.science/api/pith-number/G6C253PIK2ZU3TUZ2FPCTKW26P/events.json","paper":"https://pith.science/paper/G6C253PI"},"agent_actions":{"view_html":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P","download_json":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P.json","view_paper":"https://pith.science/paper/G6C253PI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.18872&json=true","fetch_graph":"https://pith.science/api/pith-number/G6C253PIK2ZU3TUZ2FPCTKW26P/graph.json","fetch_events":"https://pith.science/api/pith-number/G6C253PIK2ZU3TUZ2FPCTKW26P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P/action/storage_attestation","attest_author":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P/action/author_attestation","sign_citation":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P/action/citation_signature","submit_replication":"https://pith.science/pith/G6C253PIK2ZU3TUZ2FPCTKW26P/action/replication_record"}},"created_at":"2026-07-05T09:45:50.812860+00:00","updated_at":"2026-07-05T09:45:50.812860+00:00"}