{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RYPYS325MGXWG7JKFJSNVUOLSA","short_pith_number":"pith:RYPYS325","schema_version":"1.0","canonical_sha256":"8e1f896f5d61af637d2a2a64dad1cb903964fe1c09296462275d3baa4b6f7198","source":{"kind":"arxiv","id":"2507.08018","version":1},"attestation_state":"computed","paper":{"title":"Review, Remask, Refine (R3): Process-Guided Block Diffusion for Text Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Nikita Mounier, Parsa Idehpour","submitted_at":"2025-07-07T21:18:54Z","abstract_excerpt":"A key challenge for iterative text generation is enabling models to efficiently identify and correct their own errors. We propose Review, Remask, Refine (R3), a relatively simple yet elegant framework that requires no additional model training and can be applied to any pre-trained masked text diffusion model (e.g., LLaDA or BD3-LM). In R3, a Process Reward Model (PRM) is utilized for the Review of intermediate generated blocks. The framework then translates these PRM scores into a Remask strategy: the lower a block's PRM score, indicating potential mistakes, the greater the proportion of token"},"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":"2507.08018","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-07T21:18:54Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"76ff97c74f00abf1a243ea15304e047f537e5ccac734e3947af14931e4dcbd04","abstract_canon_sha256":"774d7ed50afd83716e6242764e1a613d56c81db97964484fddeae2daed17518d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:35:15.722549Z","signature_b64":"s9g4RGKEkoUpCUlwhOb5Lkmg4r2Z/2rbZbdQF/50/1MM6P5nYqTsjwRxalxcCMckhVRwE5fa57gJMlImD5VFBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8e1f896f5d61af637d2a2a64dad1cb903964fe1c09296462275d3baa4b6f7198","last_reissued_at":"2026-07-05T11:35:15.722063Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:35:15.722063Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Review, Remask, Refine (R3): Process-Guided Block Diffusion for Text Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Nikita Mounier, Parsa Idehpour","submitted_at":"2025-07-07T21:18:54Z","abstract_excerpt":"A key challenge for iterative text generation is enabling models to efficiently identify and correct their own errors. We propose Review, Remask, Refine (R3), a relatively simple yet elegant framework that requires no additional model training and can be applied to any pre-trained masked text diffusion model (e.g., LLaDA or BD3-LM). In R3, a Process Reward Model (PRM) is utilized for the Review of intermediate generated blocks. The framework then translates these PRM scores into a Remask strategy: the lower a block's PRM score, indicating potential mistakes, the greater the proportion of token"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.08018","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/2507.08018/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":"2507.08018","created_at":"2026-07-05T11:35:15.722120+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.08018v1","created_at":"2026-07-05T11:35:15.722120+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.08018","created_at":"2026-07-05T11:35:15.722120+00:00"},{"alias_kind":"pith_short_12","alias_value":"RYPYS325MGXW","created_at":"2026-07-05T11:35:15.722120+00:00"},{"alias_kind":"pith_short_16","alias_value":"RYPYS325MGXWG7JK","created_at":"2026-07-05T11:35:15.722120+00:00"},{"alias_kind":"pith_short_8","alias_value":"RYPYS325","created_at":"2026-07-05T11:35:15.722120+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21802","citing_title":"When to Plan, When to Polish: Noise Level as a Granularity Axis for Diffusion Language Models","ref_index":52,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA","json":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA.json","graph_json":"https://pith.science/api/pith-number/RYPYS325MGXWG7JKFJSNVUOLSA/graph.json","events_json":"https://pith.science/api/pith-number/RYPYS325MGXWG7JKFJSNVUOLSA/events.json","paper":"https://pith.science/paper/RYPYS325"},"agent_actions":{"view_html":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA","download_json":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA.json","view_paper":"https://pith.science/paper/RYPYS325","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.08018&json=true","fetch_graph":"https://pith.science/api/pith-number/RYPYS325MGXWG7JKFJSNVUOLSA/graph.json","fetch_events":"https://pith.science/api/pith-number/RYPYS325MGXWG7JKFJSNVUOLSA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA/action/storage_attestation","attest_author":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA/action/author_attestation","sign_citation":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA/action/citation_signature","submit_replication":"https://pith.science/pith/RYPYS325MGXWG7JKFJSNVUOLSA/action/replication_record"}},"created_at":"2026-07-05T11:35:15.722120+00:00","updated_at":"2026-07-05T11:35:15.722120+00:00"}