{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:MWMWF35W6EB7PUGEM7D64EVO2J","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"0e8deeea39032e68e54efa8dcbc3038a25f3c084329d43f21b6c526445c12029","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T13:09:49Z","title_canon_sha256":"62e201d3d163f9d7ceb6c2b5c7300780dfead246ea2ecf0c91a821de8f2a5ffb"},"schema_version":"1.0","source":{"id":"2602.18196","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.18196","created_at":"2026-05-21T01:05:16Z"},{"alias_kind":"arxiv_version","alias_value":"2602.18196v4","created_at":"2026-05-21T01:05:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.18196","created_at":"2026-05-21T01:05:16Z"},{"alias_kind":"pith_short_12","alias_value":"MWMWF35W6EB7","created_at":"2026-05-21T01:05:16Z"},{"alias_kind":"pith_short_16","alias_value":"MWMWF35W6EB7PUGE","created_at":"2026-05-21T01:05:16Z"},{"alias_kind":"pith_short_8","alias_value":"MWMWF35W","created_at":"2026-05-21T01:05:16Z"}],"graph_snapshots":[{"event_id":"sha256:eef2c1306a249847e17b80c28d6ec470a6ff296f8ddcfaee15d23ea5b32006bd","target":"graph","created_at":"2026-05-21T01:05:16Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation."}],"snapshot_sha256":"6c02010c9b714beb8dd8d24366b528de6c1b16a7bcc888b9ae1fadc04da7b1e6"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2602.18196/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"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","authors_text":"Caglar Gulcehre, Xiuying Wei","cross_cats":[],"headline":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T13:09:49Z","title":"RAT+: Train Dense, Infer Sparse -- Recurrence Augmented Attention for Dilated Inference"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.18196","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-15T20:56:31.911959Z","id":"2b217652-3625-4931-8ddd-6ce0045843c6","model_set":{"reader":"grok-4.3"},"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","pith_extraction_headline":"RAT+ lets one densely pretrained model switch to dilated sparse attention at inference with only short adaptation.","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.","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."}},"verdict_id":"2b217652-3625-4931-8ddd-6ce0045843c6"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:cfc6647bd18fd61d1ec108d68b5083382538b41952b1cea6d38836cf83d54031","target":"record","created_at":"2026-05-21T01:05:16Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"0e8deeea39032e68e54efa8dcbc3038a25f3c084329d43f21b6c526445c12029","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-02-20T13:09:49Z","title_canon_sha256":"62e201d3d163f9d7ceb6c2b5c7300780dfead246ea2ecf0c91a821de8f2a5ffb"},"schema_version":"1.0","source":{"id":"2602.18196","kind":"arxiv","version":4}},"canonical_sha256":"659962efb6f103f7d0c467c7ee12aed24291f1daa1f6c17cfa39ee8081e6df52","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"659962efb6f103f7d0c467c7ee12aed24291f1daa1f6c17cfa39ee8081e6df52","first_computed_at":"2026-05-21T01:05:16.476570Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-21T01:05:16.476570Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Xi7VHhYei70yVpFgUsyGO0oycdQBZFfSCFN8vQS8Lr4vx4MYLfK1m+yQHakrEE5Tuba9w1Tpx/g7xF0KD4F3Cg==","signature_status":"signed_v1","signed_at":"2026-05-21T01:05:16.477178Z","signed_message":"canonical_sha256_bytes"},"source_id":"2602.18196","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cfc6647bd18fd61d1ec108d68b5083382538b41952b1cea6d38836cf83d54031","sha256:eef2c1306a249847e17b80c28d6ec470a6ff296f8ddcfaee15d23ea5b32006bd"],"state_sha256":"51d399e095c7a102d403609131ff13c35cf6ae12b329f6e31753ce5d1ae70630"}