{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YKTIGY5M3L7XV5MAK5GHBVC5F6","short_pith_number":"pith:YKTIGY5M","schema_version":"1.0","canonical_sha256":"c2a68363acdaff7af580574c70d45d2fbe8298a33a41910211c16c3d6eed1cc4","source":{"kind":"arxiv","id":"2412.13328","version":2},"attestation_state":"computed","paper":{"title":"Expansion Span: Combining Fading Memory and Retrieval in Hybrid State Space Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aditya Golatkar, Benjamin Bowman, Elvis Nunez, Luca Zancato, Stefano Soatto, Wei Xia","submitted_at":"2024-12-17T20:55:42Z","abstract_excerpt":"The \"state\" of State Space Models (SSMs) represents their memory, which fades exponentially over an unbounded span. By contrast, Attention-based models have \"eidetic\" (i.e., verbatim, or photographic) memory over a finite span (context size). Hybrid architectures combine State Space layers with Attention, but still cannot recall the distant past and can access only the most recent tokens eidetically. Unlike current methods of combining SSM and Attention layers, we allow the state to be allocated based on relevancy rather than recency. In this way, for every new set of query tokens, our models "},"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":"2412.13328","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-17T20:55:42Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"38090ec6f62cc2eba206a62b2ca34e550804a60d7868446fb22c54a22b58baca","abstract_canon_sha256":"f839996ac92161b4e57e0c71c1acd70d15f7f8d9cbbd51d781a812d85d804e8f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:05.575548Z","signature_b64":"v/5sCvjb3LxHOb3359ZyEhp7lDhuFHlnEYx1W4OnHpUHv89r5p9c9lpoeXI3Z58aD8jRPvD6lRC8Z3VvZOi7Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c2a68363acdaff7af580574c70d45d2fbe8298a33a41910211c16c3d6eed1cc4","last_reissued_at":"2026-07-05T11:09:05.575025Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:05.575025Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Expansion Span: Combining Fading Memory and Retrieval in Hybrid State Space Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aditya Golatkar, Benjamin Bowman, Elvis Nunez, Luca Zancato, Stefano Soatto, Wei Xia","submitted_at":"2024-12-17T20:55:42Z","abstract_excerpt":"The \"state\" of State Space Models (SSMs) represents their memory, which fades exponentially over an unbounded span. By contrast, Attention-based models have \"eidetic\" (i.e., verbatim, or photographic) memory over a finite span (context size). Hybrid architectures combine State Space layers with Attention, but still cannot recall the distant past and can access only the most recent tokens eidetically. Unlike current methods of combining SSM and Attention layers, we allow the state to be allocated based on relevancy rather than recency. In this way, for every new set of query tokens, our models "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.13328","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/2412.13328/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":"2412.13328","created_at":"2026-07-05T11:09:05.575087+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.13328v2","created_at":"2026-07-05T11:09:05.575087+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.13328","created_at":"2026-07-05T11:09:05.575087+00:00"},{"alias_kind":"pith_short_12","alias_value":"YKTIGY5M3L7X","created_at":"2026-07-05T11:09:05.575087+00:00"},{"alias_kind":"pith_short_16","alias_value":"YKTIGY5M3L7XV5MA","created_at":"2026-07-05T11:09:05.575087+00:00"},{"alias_kind":"pith_short_8","alias_value":"YKTIGY5M","created_at":"2026-07-05T11:09:05.575087+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04378","citing_title":"DLLG: Dynamic Logit-Level Gating of LLM Experts","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2511.21016","citing_title":"Gated KalmaNet: A Fading Memory Layer Through Test-Time Ridge Regression","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6","json":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6.json","graph_json":"https://pith.science/api/pith-number/YKTIGY5M3L7XV5MAK5GHBVC5F6/graph.json","events_json":"https://pith.science/api/pith-number/YKTIGY5M3L7XV5MAK5GHBVC5F6/events.json","paper":"https://pith.science/paper/YKTIGY5M"},"agent_actions":{"view_html":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6","download_json":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6.json","view_paper":"https://pith.science/paper/YKTIGY5M","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.13328&json=true","fetch_graph":"https://pith.science/api/pith-number/YKTIGY5M3L7XV5MAK5GHBVC5F6/graph.json","fetch_events":"https://pith.science/api/pith-number/YKTIGY5M3L7XV5MAK5GHBVC5F6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6/action/storage_attestation","attest_author":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6/action/author_attestation","sign_citation":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6/action/citation_signature","submit_replication":"https://pith.science/pith/YKTIGY5M3L7XV5MAK5GHBVC5F6/action/replication_record"}},"created_at":"2026-07-05T11:09:05.575087+00:00","updated_at":"2026-07-05T11:09:05.575087+00:00"}