{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:A2M6EF3ZWKUQFSEVAUUFHSES2K","short_pith_number":"pith:A2M6EF3Z","schema_version":"1.0","canonical_sha256":"0699e21779b2a902c895052853c892d29bd24713f6ead614e8b4b2039a6d208d","source":{"kind":"arxiv","id":"2601.14758","version":4},"attestation_state":"computed","paper":{"title":"Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Hyoungjoon Lee, Injin Kong, Yohan Jo","submitted_at":"2026-01-21T08:26:51Z","abstract_excerpt":"Post-training pretrained autoregressive models (ARMs) into masked diffusion models (MDMs) has emerged as a cost-effective way to overcome the limitations of sequential generation. Yet it remains unclear whether post-trained MDMs acquire genuinely new computational mechanisms or merely re-express autoregressive computation in a non-autoregressive form. Through a comparative circuit analysis of ARMs and their MDM counterparts post-trained from the same backbones, we uncover two complementary axes of reorganization. Structurally, the shift is task-dependent: MDMs preserve autoregressive circuitry"},"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":"2601.14758","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-21T08:26:51Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"fd169fd9b0f37255df8a80a3b649e70b752113e76ec00ef4bf16b83c7c0e90f1","abstract_canon_sha256":"ad02ff691f7e418fdfb2e21840ed9f409bec2bfe8fceaae6beb39797d7c2134c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:39.617528Z","signature_b64":"qqkNP/9dGRG/MkZtN6X1hcU9zN55PMFzdtkOWOUu+Mj7wfhnkATO2zqsjR5zYd6T0L7qIz1g8YPz7tAu53tIDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0699e21779b2a902c895052853c892d29bd24713f6ead614e8b4b2039a6d208d","last_reissued_at":"2026-05-29T02:05:39.616622Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:39.616622Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Hyoungjoon Lee, Injin Kong, Yohan Jo","submitted_at":"2026-01-21T08:26:51Z","abstract_excerpt":"Post-training pretrained autoregressive models (ARMs) into masked diffusion models (MDMs) has emerged as a cost-effective way to overcome the limitations of sequential generation. Yet it remains unclear whether post-trained MDMs acquire genuinely new computational mechanisms or merely re-express autoregressive computation in a non-autoregressive form. Through a comparative circuit analysis of ARMs and their MDM counterparts post-trained from the same backbones, we uncover two complementary axes of reorganization. Structurally, the shift is task-dependent: MDMs preserve autoregressive circuitry"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.14758","kind":"arxiv","version":4},"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/2601.14758/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":"2601.14758","created_at":"2026-05-29T02:05:39.616747+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.14758v4","created_at":"2026-05-29T02:05:39.616747+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.14758","created_at":"2026-05-29T02:05:39.616747+00:00"},{"alias_kind":"pith_short_12","alias_value":"A2M6EF3ZWKUQ","created_at":"2026-05-29T02:05:39.616747+00:00"},{"alias_kind":"pith_short_16","alias_value":"A2M6EF3ZWKUQFSEV","created_at":"2026-05-29T02:05:39.616747+00:00"},{"alias_kind":"pith_short_8","alias_value":"A2M6EF3Z","created_at":"2026-05-29T02:05:39.616747+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/A2M6EF3ZWKUQFSEVAUUFHSES2K","json":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K.json","graph_json":"https://pith.science/api/pith-number/A2M6EF3ZWKUQFSEVAUUFHSES2K/graph.json","events_json":"https://pith.science/api/pith-number/A2M6EF3ZWKUQFSEVAUUFHSES2K/events.json","paper":"https://pith.science/paper/A2M6EF3Z"},"agent_actions":{"view_html":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K","download_json":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K.json","view_paper":"https://pith.science/paper/A2M6EF3Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.14758&json=true","fetch_graph":"https://pith.science/api/pith-number/A2M6EF3ZWKUQFSEVAUUFHSES2K/graph.json","fetch_events":"https://pith.science/api/pith-number/A2M6EF3ZWKUQFSEVAUUFHSES2K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K/action/storage_attestation","attest_author":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K/action/author_attestation","sign_citation":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K/action/citation_signature","submit_replication":"https://pith.science/pith/A2M6EF3ZWKUQFSEVAUUFHSES2K/action/replication_record"}},"created_at":"2026-05-29T02:05:39.616747+00:00","updated_at":"2026-05-29T02:05:39.616747+00:00"}