{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:4PR2P4ZE5G6JZ5QA2TLK5ZKRAE","short_pith_number":"pith:4PR2P4ZE","schema_version":"1.0","canonical_sha256":"e3e3a7f324e9bc9cf600d4d6aee55101195e654ae3ebcdeb280810b8b5aced0a","source":{"kind":"arxiv","id":"2207.06881","version":2},"attestation_state":"computed","paper":{"title":"Recurrent Memory Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aydar Bulatov, Mikhail S. Burtsev, Yuri Kuratov","submitted_at":"2022-07-14T13:00:22Z","abstract_excerpt":"Transformer-based models show their effectiveness across multiple domains and tasks. The self-attention allows to combine information from all sequence elements into context-aware representations. However, global and local information has to be stored mostly in the same element-wise representations. Moreover, the length of an input sequence is limited by quadratic computational complexity of self-attention.\n  In this work, we propose and study a memory-augmented segment-level recurrent Transformer (RMT). Memory allows to store and process local and global information as well as to pass informa"},"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":"2207.06881","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-07-14T13:00:22Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"6b59849f29ac63d88f4f6a7edc22f26ee4091967a752d78a9fd8010e5db37867","abstract_canon_sha256":"884d627ca4ceaac115d30c8d90ae7953106778d2c054683a43c988becff9cfcd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:23:32.338041Z","signature_b64":"Vk3yD11wf7hYqCTvd8YJV9b/yxV4gFdlSBcGhU0gsXMT7NS0iqukU8sn/eNk2jt4fTcAqNN7Jtsj2v5eJFk8Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e3e3a7f324e9bc9cf600d4d6aee55101195e654ae3ebcdeb280810b8b5aced0a","last_reissued_at":"2026-07-05T05:23:32.337585Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:23:32.337585Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Recurrent Memory Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aydar Bulatov, Mikhail S. Burtsev, Yuri Kuratov","submitted_at":"2022-07-14T13:00:22Z","abstract_excerpt":"Transformer-based models show their effectiveness across multiple domains and tasks. The self-attention allows to combine information from all sequence elements into context-aware representations. However, global and local information has to be stored mostly in the same element-wise representations. Moreover, the length of an input sequence is limited by quadratic computational complexity of self-attention.\n  In this work, we propose and study a memory-augmented segment-level recurrent Transformer (RMT). Memory allows to store and process local and global information as well as to pass informa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2207.06881","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/2207.06881/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":"2207.06881","created_at":"2026-07-05T05:23:32.337651+00:00"},{"alias_kind":"arxiv_version","alias_value":"2207.06881v2","created_at":"2026-07-05T05:23:32.337651+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2207.06881","created_at":"2026-07-05T05:23:32.337651+00:00"},{"alias_kind":"pith_short_12","alias_value":"4PR2P4ZE5G6J","created_at":"2026-07-05T05:23:32.337651+00:00"},{"alias_kind":"pith_short_16","alias_value":"4PR2P4ZE5G6JZ5QA","created_at":"2026-07-05T05:23:32.337651+00:00"},{"alias_kind":"pith_short_8","alias_value":"4PR2P4ZE","created_at":"2026-07-05T05:23:32.337651+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08032","citing_title":"What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2606.21121","citing_title":"Answer Engineering: Local Trajectory Editing for Protocol-Constrained Decision Making in Large Language Models","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20737","citing_title":"Repeated Shared Access Enables Grokking, but Edit Propagation Depends on an Addressable Memory","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24930","citing_title":"H$^{2}$MT: Semantic Hierarchy-Aware Hierarchical Memory Transformer","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27686","citing_title":"Tensor Memory: Fixed-Size Recurrent State for Long-Horizon Transformers","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22884","citing_title":"Tensor Cache: Eviction-conditioned Associative Memory for Transformers","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2508.16745","citing_title":"Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13725","citing_title":"On the Effectiveness of Context Compression for Repository-Level Tasks: An Empirical Investigation","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18580","citing_title":"Sessa: Selective State Space Attention","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21999","citing_title":"Universal Transformers Need Memory: Depth-State Trade-offs in Adaptive Recursive Reasoning","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE","json":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE.json","graph_json":"https://pith.science/api/pith-number/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/graph.json","events_json":"https://pith.science/api/pith-number/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/events.json","paper":"https://pith.science/paper/4PR2P4ZE"},"agent_actions":{"view_html":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE","download_json":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE.json","view_paper":"https://pith.science/paper/4PR2P4ZE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2207.06881&json=true","fetch_graph":"https://pith.science/api/pith-number/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/graph.json","fetch_events":"https://pith.science/api/pith-number/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/action/storage_attestation","attest_author":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/action/author_attestation","sign_citation":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/action/citation_signature","submit_replication":"https://pith.science/pith/4PR2P4ZE5G6JZ5QA2TLK5ZKRAE/action/replication_record"}},"created_at":"2026-07-05T05:23:32.337651+00:00","updated_at":"2026-07-05T05:23:32.337651+00:00"}