{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:M6WYU7X6A3YMRVUPOC7GGYCUMC","short_pith_number":"pith:M6WYU7X6","schema_version":"1.0","canonical_sha256":"67ad8a7efe06f0c8d68f70be63605460894eb794b526082c333ef1b223c29c64","source":{"kind":"arxiv","id":"2602.10314","version":2},"attestation_state":"computed","paper":{"title":"Stop Training for the Worst: Progressive Unmasking Accelerates Masked Diffusion Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"David Alvarez-Melis, Jaeyeon Kim, Jonathan Geuter, Sham Kakade, Sitan Chen","submitted_at":"2026-02-10T21:42:50Z","abstract_excerpt":"Masked Diffusion Models (MDMs) have emerged as a promising approach for generative modeling in discrete spaces. By generating sequences in any order and allowing for parallel decoding, they enable fast inference and strong performance on non-causal tasks. However, this flexibility comes with a training complexity trade-off: MDMs train on an exponentially large set of masking patterns, which is not only computationally expensive, but also creates a train--test mismatch between the random masks used in training and the highly structured masks induced by inference-time unmasking. In this work, we"},"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":"2602.10314","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-10T21:42:50Z","cross_cats_sorted":[],"title_canon_sha256":"2042d77dabd4180c254ad110f769886eca195ee05594e1b138ba995fd44c4250","abstract_canon_sha256":"63f3d3b2181ab8841d53748953afbb73bcc49cd8f19f71cf56c8f6cceebfc532"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:14:35.001000Z","signature_b64":"5P1nXpyV+zA+1uI1r71ATLaIMKU93nenTo+xf0Tjm1krjNxKNbh7M6Jk1zc9662VVCKkk6TvZ/+P4fcOGoEmCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"67ad8a7efe06f0c8d68f70be63605460894eb794b526082c333ef1b223c29c64","last_reissued_at":"2026-06-05T01:14:35.000211Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:14:35.000211Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stop Training for the Worst: Progressive Unmasking Accelerates Masked Diffusion Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"David Alvarez-Melis, Jaeyeon Kim, Jonathan Geuter, Sham Kakade, Sitan Chen","submitted_at":"2026-02-10T21:42:50Z","abstract_excerpt":"Masked Diffusion Models (MDMs) have emerged as a promising approach for generative modeling in discrete spaces. By generating sequences in any order and allowing for parallel decoding, they enable fast inference and strong performance on non-causal tasks. However, this flexibility comes with a training complexity trade-off: MDMs train on an exponentially large set of masking patterns, which is not only computationally expensive, but also creates a train--test mismatch between the random masks used in training and the highly structured masks induced by inference-time unmasking. In this work, we"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.10314","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/2602.10314/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":"2602.10314","created_at":"2026-06-05T01:14:35.000310+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.10314v2","created_at":"2026-06-05T01:14:35.000310+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.10314","created_at":"2026-06-05T01:14:35.000310+00:00"},{"alias_kind":"pith_short_12","alias_value":"M6WYU7X6A3YM","created_at":"2026-06-05T01:14:35.000310+00:00"},{"alias_kind":"pith_short_16","alias_value":"M6WYU7X6A3YMRVUP","created_at":"2026-06-05T01:14:35.000310+00:00"},{"alias_kind":"pith_short_8","alias_value":"M6WYU7X6","created_at":"2026-06-05T01:14:35.000310+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.22967","citing_title":"Learned Relay Representations for Forward-Thinking Discrete Diffusion Models","ref_index":17,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC","json":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC.json","graph_json":"https://pith.science/api/pith-number/M6WYU7X6A3YMRVUPOC7GGYCUMC/graph.json","events_json":"https://pith.science/api/pith-number/M6WYU7X6A3YMRVUPOC7GGYCUMC/events.json","paper":"https://pith.science/paper/M6WYU7X6"},"agent_actions":{"view_html":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC","download_json":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC.json","view_paper":"https://pith.science/paper/M6WYU7X6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.10314&json=true","fetch_graph":"https://pith.science/api/pith-number/M6WYU7X6A3YMRVUPOC7GGYCUMC/graph.json","fetch_events":"https://pith.science/api/pith-number/M6WYU7X6A3YMRVUPOC7GGYCUMC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC/action/storage_attestation","attest_author":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC/action/author_attestation","sign_citation":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC/action/citation_signature","submit_replication":"https://pith.science/pith/M6WYU7X6A3YMRVUPOC7GGYCUMC/action/replication_record"}},"created_at":"2026-06-05T01:14:35.000310+00:00","updated_at":"2026-06-05T01:14:35.000310+00:00"}