{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MWMWF35W6EB7PUGEM7D64EVO2J","short_pith_number":"pith:MWMWF35W","schema_version":"1.0","canonical_sha256":"659962efb6f103f7d0c467c7ee12aed24291f1daa1f6c17cfa39ee8081e6df52","source":{"kind":"arxiv","id":"2602.18196","version":4},"attestation_state":"computed","paper":{"title":"RAT+: Train Dense, Infer Sparse -- Recurrence Augmented Attention for Dilated Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Caglar Gulcehre, Xiuying Wei","submitted_at":"2026-02-20T13:09:49Z","abstract_excerpt":"Structured dilated attention has an appealing inference-time efficiency knob: it reduces the FLOPs of attention and the KV cache size by a factor of the dilation size D, while preserving long-range connectivity. While prior work studies it by training each configuration from scratch, directly sparsifying a pretrained attention model into a dilated pattern leads to severe accuracy degradation, preventing flexible reuse across inference scenarios. We introduce RAT+, a dense-pretraining architecture that augments attention with full-sequence recurrence and active recurrence learning. A single RAT"},"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":"2602.18196","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T13:09:49Z","cross_cats_sorted":[],"title_canon_sha256":"62e201d3d163f9d7ceb6c2b5c7300780dfead246ea2ecf0c91a821de8f2a5ffb","abstract_canon_sha256":"0e8deeea39032e68e54efa8dcbc3038a25f3c084329d43f21b6c526445c12029"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:16.477178Z","signature_b64":"Xi7VHhYei70yVpFgUsyGO0oycdQBZFfSCFN8vQS8Lr4vx4MYLfK1m+yQHakrEE5Tuba9w1Tpx/g7xF0KD4F3Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"659962efb6f103f7d0c467c7ee12aed24291f1daa1f6c17cfa39ee8081e6df52","last_reissued_at":"2026-05-21T01:05:16.476570Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:16.476570Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RAT+: Train Dense, Infer Sparse -- Recurrence Augmented Attention for Dilated Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Caglar Gulcehre, Xiuying Wei","submitted_at":"2026-02-20T13:09:49Z","abstract_excerpt":"Structured dilated attention has an appealing inference-time efficiency knob: it reduces the FLOPs of attention and the KV cache size by a factor of the dilation size D, while preserving long-range connectivity. While prior work studies it by training each configuration from scratch, directly sparsifying a pretrained attention model into a dilated pattern leads to severe accuracy degradation, preventing flexible reuse across inference scenarios. We introduce RAT+, a dense-pretraining architecture that augments attention with full-sequence recurrence and active recurrence learning. A single RAT"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"A single RAT+ model is pretrained densely once and can then be flexibly switched at inference time to dilated attention (optionally with local windows) or hybrid layer/head compositions, requiring only a short 1B-token resolution adaptation rather than retraining separate sparse models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That adding full-sequence recurrence and active recurrence learning during dense pretraining creates representations that transfer to dilated sparse patterns with only short adaptation and limited accuracy loss.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RAT+ pretrains a single dense recurrent-augmented attention model that supports flexible dilated sparse inference after short adaptation, matching dense accuracy at moderate dilation and losing only 1-3 points at high dilation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"6c02010c9b714beb8dd8d24366b528de6c1b16a7bcc888b9ae1fadc04da7b1e6"},"source":{"id":"2602.18196","kind":"arxiv","version":4},"verdict":{"id":"2b217652-3625-4931-8ddd-6ce0045843c6","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T20:56:31.911959Z","strongest_claim":"A single RAT+ model is pretrained densely once and can then be flexibly switched at inference time to dilated attention (optionally with local windows) or hybrid layer/head compositions, requiring only a short 1B-token resolution adaptation rather than retraining separate sparse models.","one_line_summary":"RAT+ pretrains a single dense recurrent-augmented attention model that supports flexible dilated sparse inference after short adaptation, matching dense accuracy at moderate dilation and losing only 1-3 points at high dilation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That adding full-sequence recurrence and active recurrence learning during dense pretraining creates representations that transfer to dilated sparse patterns with only short adaptation and limited accuracy loss.","pith_extraction_headline":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2602.18196/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":"2602.18196","created_at":"2026-05-21T01:05:16.476635+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.18196v4","created_at":"2026-05-21T01:05:16.476635+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.18196","created_at":"2026-05-21T01:05:16.476635+00:00"},{"alias_kind":"pith_short_12","alias_value":"MWMWF35W6EB7","created_at":"2026-05-21T01:05:16.476635+00:00"},{"alias_kind":"pith_short_16","alias_value":"MWMWF35W6EB7PUGE","created_at":"2026-05-21T01:05:16.476635+00:00"},{"alias_kind":"pith_short_8","alias_value":"MWMWF35W","created_at":"2026-05-21T01:05:16.476635+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/MWMWF35W6EB7PUGEM7D64EVO2J","json":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J.json","graph_json":"https://pith.science/api/pith-number/MWMWF35W6EB7PUGEM7D64EVO2J/graph.json","events_json":"https://pith.science/api/pith-number/MWMWF35W6EB7PUGEM7D64EVO2J/events.json","paper":"https://pith.science/paper/MWMWF35W"},"agent_actions":{"view_html":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J","download_json":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J.json","view_paper":"https://pith.science/paper/MWMWF35W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.18196&json=true","fetch_graph":"https://pith.science/api/pith-number/MWMWF35W6EB7PUGEM7D64EVO2J/graph.json","fetch_events":"https://pith.science/api/pith-number/MWMWF35W6EB7PUGEM7D64EVO2J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J/action/storage_attestation","attest_author":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J/action/author_attestation","sign_citation":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J/action/citation_signature","submit_replication":"https://pith.science/pith/MWMWF35W6EB7PUGEM7D64EVO2J/action/replication_record"}},"created_at":"2026-05-21T01:05:16.476635+00:00","updated_at":"2026-05-21T01:05:16.476635+00:00"}