{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:5OYAESIRMYPP7XHXLNTNRWRQNS","short_pith_number":"pith:5OYAESIR","schema_version":"1.0","canonical_sha256":"ebb0024911661effdcf75b66d8da306ca84c4e61afb2ce70c5b09c11c120f589","source":{"kind":"arxiv","id":"2506.12035","version":1},"attestation_state":"computed","paper":{"title":"MARch\\'e: Fast Masked Autoregressive Image Generation with Cache-Aware Attention","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chaoyi Jiang, Hossein Entezari Zarch, Lei Gao, Murali Annavaram, Sungwoo Kim, Won Woo Ro","submitted_at":"2025-05-22T23:26:56Z","abstract_excerpt":"Masked autoregressive (MAR) models unify the strengths of masked and autoregressive generation by predicting tokens in a fixed order using bidirectional attention for image generation. While effective, MAR models suffer from significant computational overhead, as they recompute attention and feed-forward representations for all tokens at every decoding step, despite most tokens remaining semantically stable across steps. We propose a training-free generation framework MARch\\'e to address this inefficiency through two key components: cache-aware attention and selective KV refresh. Cache-aware a"},"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":"2506.12035","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-22T23:26:56Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"21776bab72c1dbadc31e4a8c32780e9bacaae78aced0aa09c578768724d5d167","abstract_canon_sha256":"9b4019a6ae846e6ba32b164c80a31846e5868245873513660a271dc063fae82c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:21:15.993583Z","signature_b64":"acZkG7wpoIh8O+5Q0EoyhrFSLB23ZsfPKZcLDsS31fSyFI0oaJoy73EJtfQaSIMAczus/UrOU94H1DiLczSFCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ebb0024911661effdcf75b66d8da306ca84c4e61afb2ce70c5b09c11c120f589","last_reissued_at":"2026-07-05T11:21:15.993009Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:21:15.993009Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MARch\\'e: Fast Masked Autoregressive Image Generation with Cache-Aware Attention","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chaoyi Jiang, Hossein Entezari Zarch, Lei Gao, Murali Annavaram, Sungwoo Kim, Won Woo Ro","submitted_at":"2025-05-22T23:26:56Z","abstract_excerpt":"Masked autoregressive (MAR) models unify the strengths of masked and autoregressive generation by predicting tokens in a fixed order using bidirectional attention for image generation. While effective, MAR models suffer from significant computational overhead, as they recompute attention and feed-forward representations for all tokens at every decoding step, despite most tokens remaining semantically stable across steps. We propose a training-free generation framework MARch\\'e to address this inefficiency through two key components: cache-aware attention and selective KV refresh. Cache-aware a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.12035","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/2506.12035/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":"2506.12035","created_at":"2026-07-05T11:21:15.993091+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.12035v1","created_at":"2026-07-05T11:21:15.993091+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.12035","created_at":"2026-07-05T11:21:15.993091+00:00"},{"alias_kind":"pith_short_12","alias_value":"5OYAESIRMYPP","created_at":"2026-07-05T11:21:15.993091+00:00"},{"alias_kind":"pith_short_16","alias_value":"5OYAESIRMYPP7XHX","created_at":"2026-07-05T11:21:15.993091+00:00"},{"alias_kind":"pith_short_8","alias_value":"5OYAESIR","created_at":"2026-07-05T11:21:15.993091+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/5OYAESIRMYPP7XHXLNTNRWRQNS","json":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS.json","graph_json":"https://pith.science/api/pith-number/5OYAESIRMYPP7XHXLNTNRWRQNS/graph.json","events_json":"https://pith.science/api/pith-number/5OYAESIRMYPP7XHXLNTNRWRQNS/events.json","paper":"https://pith.science/paper/5OYAESIR"},"agent_actions":{"view_html":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS","download_json":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS.json","view_paper":"https://pith.science/paper/5OYAESIR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.12035&json=true","fetch_graph":"https://pith.science/api/pith-number/5OYAESIRMYPP7XHXLNTNRWRQNS/graph.json","fetch_events":"https://pith.science/api/pith-number/5OYAESIRMYPP7XHXLNTNRWRQNS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS/action/storage_attestation","attest_author":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS/action/author_attestation","sign_citation":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS/action/citation_signature","submit_replication":"https://pith.science/pith/5OYAESIRMYPP7XHXLNTNRWRQNS/action/replication_record"}},"created_at":"2026-07-05T11:21:15.993091+00:00","updated_at":"2026-07-05T11:21:15.993091+00:00"}