{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DFH3FVJX5MGSSZCSWYYR4AC5ZO","short_pith_number":"pith:DFH3FVJX","schema_version":"1.0","canonical_sha256":"194fb2d537eb0d296452b6311e005dcb93ab210684c46e77e4372bc053037129","source":{"kind":"arxiv","id":"2601.21686","version":2},"attestation_state":"computed","paper":{"title":"Don't be so Stief! Learning KV Cache low-rank approximation over the Stiefel manifold","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Ahmet Caner Y\\\"uz\\\"ug\\\"uler, Alessio Burrello, Andrea Vannozzi, Daniele Jahier Pagliari, Enrico Macii, Luca Benfenati, Lukas Cavigelli, Matteo Risso","submitted_at":"2026-01-29T13:19:24Z","abstract_excerpt":"Key-value (KV) caching enables fast autoregressive decoding but at long contexts becomes a dominant bottleneck in High Bandwidth Memory (HBM) capacity and bandwidth. A common mitigation is to compress cached keys and values by projecting per-head matrices to a lower rank, storing only the projections in the HBM. However, existing post-training approaches typically fit these projections using SVD-style proxy objectives, which may poorly reflect end-to-end reconstruction after softmax, value mixing, and subsequent decoder-layer transformations.\n  For these reasons, we introduce StiefAttention, a"},"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":"2601.21686","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-29T13:19:24Z","cross_cats_sorted":[],"title_canon_sha256":"f20a1e72d517850567ef90c38b1bb5e86ca42668226a82a164db47e14c577e69","abstract_canon_sha256":"2c5c9b737d4a9a3e1c9d1fad02654081b0b49ab981ae860f6e0e131013a27df0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-01T01:03:48.783481Z","signature_b64":"vSgASfBKdYvEypDpk3NcijDUA9ndBikTM2EcsZHPodSWrNz7EoTJzd8qysH7Dwdfzq7TLr0Jts37FRmE3oAqAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"194fb2d537eb0d296452b6311e005dcb93ab210684c46e77e4372bc053037129","last_reissued_at":"2026-06-01T01:03:48.782405Z","signature_status":"signed_v1","first_computed_at":"2026-06-01T01:03:48.782405Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Don't be so Stief! Learning KV Cache low-rank approximation over the Stiefel manifold","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Ahmet Caner Y\\\"uz\\\"ug\\\"uler, Alessio Burrello, Andrea Vannozzi, Daniele Jahier Pagliari, Enrico Macii, Luca Benfenati, Lukas Cavigelli, Matteo Risso","submitted_at":"2026-01-29T13:19:24Z","abstract_excerpt":"Key-value (KV) caching enables fast autoregressive decoding but at long contexts becomes a dominant bottleneck in High Bandwidth Memory (HBM) capacity and bandwidth. A common mitigation is to compress cached keys and values by projecting per-head matrices to a lower rank, storing only the projections in the HBM. However, existing post-training approaches typically fit these projections using SVD-style proxy objectives, which may poorly reflect end-to-end reconstruction after softmax, value mixing, and subsequent decoder-layer transformations.\n  For these reasons, we introduce StiefAttention, a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.21686","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/2601.21686/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":"2601.21686","created_at":"2026-06-01T01:03:48.782540+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.21686v2","created_at":"2026-06-01T01:03:48.782540+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.21686","created_at":"2026-06-01T01:03:48.782540+00:00"},{"alias_kind":"pith_short_12","alias_value":"DFH3FVJX5MGS","created_at":"2026-06-01T01:03:48.782540+00:00"},{"alias_kind":"pith_short_16","alias_value":"DFH3FVJX5MGSSZCS","created_at":"2026-06-01T01:03:48.782540+00:00"},{"alias_kind":"pith_short_8","alias_value":"DFH3FVJX","created_at":"2026-06-01T01:03:48.782540+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/DFH3FVJX5MGSSZCSWYYR4AC5ZO","json":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO.json","graph_json":"https://pith.science/api/pith-number/DFH3FVJX5MGSSZCSWYYR4AC5ZO/graph.json","events_json":"https://pith.science/api/pith-number/DFH3FVJX5MGSSZCSWYYR4AC5ZO/events.json","paper":"https://pith.science/paper/DFH3FVJX"},"agent_actions":{"view_html":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO","download_json":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO.json","view_paper":"https://pith.science/paper/DFH3FVJX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.21686&json=true","fetch_graph":"https://pith.science/api/pith-number/DFH3FVJX5MGSSZCSWYYR4AC5ZO/graph.json","fetch_events":"https://pith.science/api/pith-number/DFH3FVJX5MGSSZCSWYYR4AC5ZO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO/action/storage_attestation","attest_author":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO/action/author_attestation","sign_citation":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO/action/citation_signature","submit_replication":"https://pith.science/pith/DFH3FVJX5MGSSZCSWYYR4AC5ZO/action/replication_record"}},"created_at":"2026-06-01T01:03:48.782540+00:00","updated_at":"2026-06-01T01:03:48.782540+00:00"}