{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5JQ7EVSVRIKBVZET4VKNKPYPDU","short_pith_number":"pith:5JQ7EVSV","schema_version":"1.0","canonical_sha256":"ea61f256558a141ae493e554d53f0f1d1ceb969dbcc140ad0fff1e2fe6d93b88","source":{"kind":"arxiv","id":"2605.27686","version":1},"attestation_state":"computed","paper":{"title":"Tensor Memory: Fixed-Size Recurrent State for Long-Horizon Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Antonio Torralba, Daniel Karl I. Weidele, Kabir Swain, Mauro Martino, Sijie Han","submitted_at":"2026-05-26T21:03:42Z","abstract_excerpt":"Transformers process images and videos by flattening space and time into long token sequences. While attention and KV caching preserve past features, their memory grows with sequence length and they lack an explicit, persistent spatial state, making long-horizon video understanding and occlusion-sensitive reasoning difficult. We propose Tensor Memory, a lightweight module that augments Transformer blocks with a fixed-size recurrent 3D memory tensor: tokens write into a voxel grid via a differentiable soft write that deposits content as a Gaussian-weighted volume around a predicted continuous 3"},"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":"2605.27686","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-26T21:03:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"73f2ba2bb169a14afb5cc58118555ff5dc7d914b1557102c96522828056b07df","abstract_canon_sha256":"dab2c8e6da785780ae3934ad15133ee225fb6705b7a4883a97dc006f36309a2f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T01:04:46.201364Z","signature_b64":"+m3DO4NkzOd39LnwPStBLNTXCLuaNTHHupi6YHoNPqu370AeCRKTINtTYeUYzHQa1LTTLbZppENq1zZvbkbsBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ea61f256558a141ae493e554d53f0f1d1ceb969dbcc140ad0fff1e2fe6d93b88","last_reissued_at":"2026-05-28T01:04:46.200933Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T01:04:46.200933Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tensor Memory: Fixed-Size Recurrent State for Long-Horizon Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Antonio Torralba, Daniel Karl I. Weidele, Kabir Swain, Mauro Martino, Sijie Han","submitted_at":"2026-05-26T21:03:42Z","abstract_excerpt":"Transformers process images and videos by flattening space and time into long token sequences. While attention and KV caching preserve past features, their memory grows with sequence length and they lack an explicit, persistent spatial state, making long-horizon video understanding and occlusion-sensitive reasoning difficult. We propose Tensor Memory, a lightweight module that augments Transformer blocks with a fixed-size recurrent 3D memory tensor: tokens write into a voxel grid via a differentiable soft write that deposits content as a Gaussian-weighted volume around a predicted continuous 3"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.27686","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/2605.27686/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":"2605.27686","created_at":"2026-05-28T01:04:46.200991+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.27686v1","created_at":"2026-05-28T01:04:46.200991+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.27686","created_at":"2026-05-28T01:04:46.200991+00:00"},{"alias_kind":"pith_short_12","alias_value":"5JQ7EVSVRIKB","created_at":"2026-05-28T01:04:46.200991+00:00"},{"alias_kind":"pith_short_16","alias_value":"5JQ7EVSVRIKBVZET","created_at":"2026-05-28T01:04:46.200991+00:00"},{"alias_kind":"pith_short_8","alias_value":"5JQ7EVSV","created_at":"2026-05-28T01:04:46.200991+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/5JQ7EVSVRIKBVZET4VKNKPYPDU","json":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU.json","graph_json":"https://pith.science/api/pith-number/5JQ7EVSVRIKBVZET4VKNKPYPDU/graph.json","events_json":"https://pith.science/api/pith-number/5JQ7EVSVRIKBVZET4VKNKPYPDU/events.json","paper":"https://pith.science/paper/5JQ7EVSV"},"agent_actions":{"view_html":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU","download_json":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU.json","view_paper":"https://pith.science/paper/5JQ7EVSV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.27686&json=true","fetch_graph":"https://pith.science/api/pith-number/5JQ7EVSVRIKBVZET4VKNKPYPDU/graph.json","fetch_events":"https://pith.science/api/pith-number/5JQ7EVSVRIKBVZET4VKNKPYPDU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU/action/storage_attestation","attest_author":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU/action/author_attestation","sign_citation":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU/action/citation_signature","submit_replication":"https://pith.science/pith/5JQ7EVSVRIKBVZET4VKNKPYPDU/action/replication_record"}},"created_at":"2026-05-28T01:04:46.200991+00:00","updated_at":"2026-05-28T01:04:46.200991+00:00"}