{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FGXTZGOA4LJPGD2ADHQJQTB4Y4","short_pith_number":"pith:FGXTZGOA","schema_version":"1.0","canonical_sha256":"29af3c99c0e2d2f30f4019e0984c3cc73694939edc7ccdc94fe21f09fad9df00","source":{"kind":"arxiv","id":"2605.30991","version":1},"attestation_state":"computed","paper":{"title":"Parallel Tempering Initial Sampling in Inference-Time Reward Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Gwangho Kim, Myeongjun Oh, Sungyoon Lee","submitted_at":"2026-05-29T08:26:14Z","abstract_excerpt":"Inference-time reward alignment steers pretrained diffusion and flow-based generative models to satisfy user-specified rewards without retraining. Recently, Sequential Monte Carlo (SMC) has emerged as a powerful framework for this task by iteratively filtering and propagating multiple particles. However, we show that standard SMC-based methods often suffer from poor performance because they initialize particles from a standard prior, whereas high-reward regions in complex reward landscapes are extremely rare. Further, we show that even recent reward-aware initial sampling approaches remain vul"},"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":"2605.30991","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T08:26:14Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"2dce1e286dfb3d8389897ea489b1bfddeb74ba72e9b8327dc553cd6f753d1069","abstract_canon_sha256":"ec8a4e5b898bf6963f96246bf091739eda5b9e6fd08c55bd5d25945155efd7d4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-01T01:03:29.160874Z","signature_b64":"uT7wCSgStDpi+lNcl25T9uhsU82CyHHOOHcfHjoloppu2X+R6dAyG8V6pnjuowlLiS0jDoJR1dlxRq7h01JnDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"29af3c99c0e2d2f30f4019e0984c3cc73694939edc7ccdc94fe21f09fad9df00","last_reissued_at":"2026-06-01T01:03:29.159998Z","signature_status":"signed_v1","first_computed_at":"2026-06-01T01:03:29.159998Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Parallel Tempering Initial Sampling in Inference-Time Reward Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Gwangho Kim, Myeongjun Oh, Sungyoon Lee","submitted_at":"2026-05-29T08:26:14Z","abstract_excerpt":"Inference-time reward alignment steers pretrained diffusion and flow-based generative models to satisfy user-specified rewards without retraining. Recently, Sequential Monte Carlo (SMC) has emerged as a powerful framework for this task by iteratively filtering and propagating multiple particles. However, we show that standard SMC-based methods often suffer from poor performance because they initialize particles from a standard prior, whereas high-reward regions in complex reward landscapes are extremely rare. Further, we show that even recent reward-aware initial sampling approaches remain vul"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.30991","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/2605.30991/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":"2605.30991","created_at":"2026-06-01T01:03:29.160150+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.30991v1","created_at":"2026-06-01T01:03:29.160150+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.30991","created_at":"2026-06-01T01:03:29.160150+00:00"},{"alias_kind":"pith_short_12","alias_value":"FGXTZGOA4LJP","created_at":"2026-06-01T01:03:29.160150+00:00"},{"alias_kind":"pith_short_16","alias_value":"FGXTZGOA4LJPGD2A","created_at":"2026-06-01T01:03:29.160150+00:00"},{"alias_kind":"pith_short_8","alias_value":"FGXTZGOA","created_at":"2026-06-01T01:03:29.160150+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/FGXTZGOA4LJPGD2ADHQJQTB4Y4","json":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4.json","graph_json":"https://pith.science/api/pith-number/FGXTZGOA4LJPGD2ADHQJQTB4Y4/graph.json","events_json":"https://pith.science/api/pith-number/FGXTZGOA4LJPGD2ADHQJQTB4Y4/events.json","paper":"https://pith.science/paper/FGXTZGOA"},"agent_actions":{"view_html":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4","download_json":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4.json","view_paper":"https://pith.science/paper/FGXTZGOA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.30991&json=true","fetch_graph":"https://pith.science/api/pith-number/FGXTZGOA4LJPGD2ADHQJQTB4Y4/graph.json","fetch_events":"https://pith.science/api/pith-number/FGXTZGOA4LJPGD2ADHQJQTB4Y4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4/action/storage_attestation","attest_author":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4/action/author_attestation","sign_citation":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4/action/citation_signature","submit_replication":"https://pith.science/pith/FGXTZGOA4LJPGD2ADHQJQTB4Y4/action/replication_record"}},"created_at":"2026-06-01T01:03:29.160150+00:00","updated_at":"2026-06-01T01:03:29.160150+00:00"}