{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2PW4MWMSGOIUKMR7PE636VL4UZ","short_pith_number":"pith:2PW4MWMS","schema_version":"1.0","canonical_sha256":"d3edc65992339145323f793dbf557ca66cfc6d6c45064e79159fee58e24b3028","source":{"kind":"arxiv","id":"2412.05334","version":2},"attestation_state":"computed","paper":{"title":"Closed-Loop Supervised Fine-Tuning of Tokenized Traffic Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Boris Ivanovic, Marco Pavone, Maximilian Igl, Peter Karkus, Wenhao Ding, Yuxiao Chen, Zhejun Zhang","submitted_at":"2024-12-05T21:00:21Z","abstract_excerpt":"Traffic simulation aims to learn a policy for traffic agents that, when unrolled in closed-loop, faithfully recovers the joint distribution of trajectories observed in the real world. Inspired by large language models, tokenized multi-agent policies have recently become the state-of-the-art in traffic simulation. However, they are typically trained through open-loop behavior cloning, and thus suffer from covariate shift when executed in closed-loop during simulation. In this work, we present Closest Among Top-K (CAT-K) rollouts, a simple yet effective closed-loop fine-tuning strategy to mitiga"},"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":"2412.05334","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-05T21:00:21Z","cross_cats_sorted":[],"title_canon_sha256":"e56a190c491f241808721782b45960deb92edd80caf25a89c08626d48905a4d6","abstract_canon_sha256":"c768e39e74bb28446311f693e5920e5b49ad7d025dbe126b0817c908f18ce159"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:30:59.114452Z","signature_b64":"7ZlEsQG1ZILLq0VtuWOIcORPBas3KiqarcCPuLAqms27fpcTFmWuujLTUmuttJMlvgZCtiKLufAuDh3zcjrRCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d3edc65992339145323f793dbf557ca66cfc6d6c45064e79159fee58e24b3028","last_reissued_at":"2026-07-05T10:30:59.113755Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:30:59.113755Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Closed-Loop Supervised Fine-Tuning of Tokenized Traffic Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Boris Ivanovic, Marco Pavone, Maximilian Igl, Peter Karkus, Wenhao Ding, Yuxiao Chen, Zhejun Zhang","submitted_at":"2024-12-05T21:00:21Z","abstract_excerpt":"Traffic simulation aims to learn a policy for traffic agents that, when unrolled in closed-loop, faithfully recovers the joint distribution of trajectories observed in the real world. Inspired by large language models, tokenized multi-agent policies have recently become the state-of-the-art in traffic simulation. However, they are typically trained through open-loop behavior cloning, and thus suffer from covariate shift when executed in closed-loop during simulation. In this work, we present Closest Among Top-K (CAT-K) rollouts, a simple yet effective closed-loop fine-tuning strategy to mitiga"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.05334","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/2412.05334/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":"2412.05334","created_at":"2026-07-05T10:30:59.113835+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.05334v2","created_at":"2026-07-05T10:30:59.113835+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.05334","created_at":"2026-07-05T10:30:59.113835+00:00"},{"alias_kind":"pith_short_12","alias_value":"2PW4MWMSGOIU","created_at":"2026-07-05T10:30:59.113835+00:00"},{"alias_kind":"pith_short_16","alias_value":"2PW4MWMSGOIUKMR7","created_at":"2026-07-05T10:30:59.113835+00:00"},{"alias_kind":"pith_short_8","alias_value":"2PW4MWMS","created_at":"2026-07-05T10:30:59.113835+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.13757","citing_title":"AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10564","citing_title":"DeepSight: Long-Horizon World Modeling via Latent States Prediction for End-to-End Autonomous Driving","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ","json":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ.json","graph_json":"https://pith.science/api/pith-number/2PW4MWMSGOIUKMR7PE636VL4UZ/graph.json","events_json":"https://pith.science/api/pith-number/2PW4MWMSGOIUKMR7PE636VL4UZ/events.json","paper":"https://pith.science/paper/2PW4MWMS"},"agent_actions":{"view_html":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ","download_json":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ.json","view_paper":"https://pith.science/paper/2PW4MWMS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.05334&json=true","fetch_graph":"https://pith.science/api/pith-number/2PW4MWMSGOIUKMR7PE636VL4UZ/graph.json","fetch_events":"https://pith.science/api/pith-number/2PW4MWMSGOIUKMR7PE636VL4UZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ/action/storage_attestation","attest_author":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ/action/author_attestation","sign_citation":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ/action/citation_signature","submit_replication":"https://pith.science/pith/2PW4MWMSGOIUKMR7PE636VL4UZ/action/replication_record"}},"created_at":"2026-07-05T10:30:59.113835+00:00","updated_at":"2026-07-05T10:30:59.113835+00:00"}