{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:W4NCQJXFQX6A3N2ABDP66LOJU6","short_pith_number":"pith:W4NCQJXF","schema_version":"1.0","canonical_sha256":"b71a2826e585fc0db74008dfef2dc9a7b198a92d1e7d4c67168d3feb30d925a5","source":{"kind":"arxiv","id":"2404.07143","version":2},"attestation_state":"computed","paper":{"title":"Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.NE"],"primary_cat":"cs.CL","authors_text":"Manaal Faruqui, Siddharth Gopal, Tsendsuren Munkhdalai","submitted_at":"2024-04-10T16:18:42Z","abstract_excerpt":"This work introduces an efficient method to scale Transformer-based Large Language Models (LLMs) to infinitely long inputs with bounded memory and computation. A key component in our proposed approach is a new attention technique dubbed Infini-attention. The Infini-attention incorporates a compressive memory into the vanilla attention mechanism and builds in both masked local attention and long-term linear attention mechanisms in a single Transformer block. We demonstrate the effectiveness of our approach on long-context language modeling benchmarks, 1M sequence length passkey context block re"},"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":"2404.07143","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-10T16:18:42Z","cross_cats_sorted":["cs.AI","cs.LG","cs.NE"],"title_canon_sha256":"942d5e2c697f94cf314e8c2bf2d8b78af09d789fc1efa1f76e02432c19485d83","abstract_canon_sha256":"b0d1d87dcfe42a4e0194d4580a5947b63405af4f1fb3135e5f486eeace5cc186"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T18:09:39.831171Z","signature_b64":"A0bq+pxsrrKzfH1FdnxADx+/O1LYqMPwuzyTe3yMh7EPs9ATbrIjbz92hF80fQRWZOQuMWvJbSwR9NOaQzGrBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b71a2826e585fc0db74008dfef2dc9a7b198a92d1e7d4c67168d3feb30d925a5","last_reissued_at":"2026-05-21T18:09:39.829270Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T18:09:39.829270Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.NE"],"primary_cat":"cs.CL","authors_text":"Manaal Faruqui, Siddharth Gopal, Tsendsuren Munkhdalai","submitted_at":"2024-04-10T16:18:42Z","abstract_excerpt":"This work introduces an efficient method to scale Transformer-based Large Language Models (LLMs) to infinitely long inputs with bounded memory and computation. A key component in our proposed approach is a new attention technique dubbed Infini-attention. The Infini-attention incorporates a compressive memory into the vanilla attention mechanism and builds in both masked local attention and long-term linear attention mechanisms in a single Transformer block. We demonstrate the effectiveness of our approach on long-context language modeling benchmarks, 1M sequence length passkey context block re"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.07143","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/2404.07143/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":"2404.07143","created_at":"2026-05-21T18:09:39.829376+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.07143v2","created_at":"2026-05-21T18:09:39.829376+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.07143","created_at":"2026-05-21T18:09:39.829376+00:00"},{"alias_kind":"pith_short_12","alias_value":"W4NCQJXFQX6A","created_at":"2026-05-21T18:09:39.829376+00:00"},{"alias_kind":"pith_short_16","alias_value":"W4NCQJXFQX6A3N2A","created_at":"2026-05-21T18:09:39.829376+00:00"},{"alias_kind":"pith_short_8","alias_value":"W4NCQJXF","created_at":"2026-05-21T18:09:39.829376+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":23,"internal_anchor_count":23,"sample":[{"citing_arxiv_id":"2501.19201","citing_title":"Efficient Reasoning with Hidden Thinking","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2511.21016","citing_title":"Gated KalmaNet: A Fading Memory Layer Through Test-Time Ridge Regression","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16366","citing_title":"Fre-Res: Frequency-Residual Video Token Compression for Efficient Video MLLMs","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17410","citing_title":"Computational Challenges in Token Economics: Bridging Economic Theory and AI System Design","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19020","citing_title":"A Systematic Failure Analysis of Vision Foundation Models for Open Set Iris Presentation Attack Detection","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2506.13674","citing_title":"PrefixMemory-Tuning: Modernizing Prefix-Tuning by Decoupling the Prefix from Attention","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2507.09025","citing_title":"Lizard: An Efficient Linearization Framework for Large Language Models","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2601.11913","citing_title":"LSTM-MAS: A Long Short-Term Memory Inspired Multi-Agent System for Long-Context Understanding","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2602.13933","citing_title":"HyMem: Hybrid Memory Architecture with Dynamic Retrieval Scheduling","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2507.07957","citing_title":"MIRIX: Multi-Agent Memory System for LLM-Based Agents","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2501.00663","citing_title":"Titans: Learning to Memorize at Test Time","ref_index":73,"is_internal_anchor":true},{"citing_arxiv_id":"2510.26692","citing_title":"Kimi Linear: An Expressive, Efficient Attention Architecture","ref_index":68,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12471","citing_title":"KV-Fold: One-Step KV-Cache Recurrence for Long-Context Inference","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11051","citing_title":"On Problems of Implicit Context Compression for Software Engineering Agents","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08301","citing_title":"Priming: Hybrid State Space Models From Pre-trained Transformers","ref_index":60,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10544","citing_title":"Where Does Long-Context Supervision Actually Go? Effective-Context Exposure Balancing","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10643","citing_title":"A Single-Layer Model Can Do Language Modeling","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10268","citing_title":"MemReread: Enhancing Agentic Long-Context Reasoning via Memory-Guided Rereading","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06473","citing_title":"MICA: Multivariate Infini Compressive Attention for Time Series Forecasting","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05066","citing_title":"The Impossibility Triangle of Long-Context Modeling","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2511.16624","citing_title":"SAM 3D: 3Dfy Anything in Images","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06473","citing_title":"MICA: Multivariate Infini Compressive Attention for Time Series Forecasting","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21999","citing_title":"Universal Transformers Need Memory: Depth-State Trade-offs in Adaptive Recursive Reasoning","ref_index":21,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6","json":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6.json","graph_json":"https://pith.science/api/pith-number/W4NCQJXFQX6A3N2ABDP66LOJU6/graph.json","events_json":"https://pith.science/api/pith-number/W4NCQJXFQX6A3N2ABDP66LOJU6/events.json","paper":"https://pith.science/paper/W4NCQJXF"},"agent_actions":{"view_html":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6","download_json":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6.json","view_paper":"https://pith.science/paper/W4NCQJXF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.07143&json=true","fetch_graph":"https://pith.science/api/pith-number/W4NCQJXFQX6A3N2ABDP66LOJU6/graph.json","fetch_events":"https://pith.science/api/pith-number/W4NCQJXFQX6A3N2ABDP66LOJU6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6/action/storage_attestation","attest_author":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6/action/author_attestation","sign_citation":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6/action/citation_signature","submit_replication":"https://pith.science/pith/W4NCQJXFQX6A3N2ABDP66LOJU6/action/replication_record"}},"created_at":"2026-05-21T18:09:39.829376+00:00","updated_at":"2026-05-21T18:09:39.829376+00:00"}