{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4FBCUMWK6YCRT7SHBFVRD2ZN5T","short_pith_number":"pith:4FBCUMWK","schema_version":"1.0","canonical_sha256":"e1422a32caf60519fe47096b11eb2decf3f44d490f902cb203f3f83e834a45e2","source":{"kind":"arxiv","id":"2401.03462","version":3},"attestation_state":"computed","paper":{"title":"Long Context Compression with Activation Beacon","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ninglu Shao, Peitian Zhang, Qiwei Ye, Shitao Xiao, Zheng Liu, Zhicheng Dou","submitted_at":"2024-01-07T11:57:40Z","abstract_excerpt":"Long context compression is a critical research problem due to its significance in reducing the high computational and memory costs associated with LLMs. In this paper, we propose Activation Beacon, a plug-in module for transformer-based LLMs that targets effective, efficient, and flexible compression of long contexts. To achieve this, our method introduces the following technical designs. 1) We directly compress the activations (i.e. keys and values at every layer), rather than leveraging soft prompts to relay information (which constitute a major bottleneck to encapsulate the complex 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":"2401.03462","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-07T11:57:40Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c95fc1a6d966a6a255d925a69597cdfd492191239db17605fe47898419f20c81","abstract_canon_sha256":"5fe99ee81c34fd740ff22d22e783d29781c6b64c095cef34acf7214e562c0153"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:18:51.706884Z","signature_b64":"iYa0mnOrdtFkxTE45P/+Anw4zQ+r/7TXuhnKLLMq9naVohqgXWx3Efr5u5Lm/YBgPhMj/uO+eGnZ4BVfQyGOBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e1422a32caf60519fe47096b11eb2decf3f44d490f902cb203f3f83e834a45e2","last_reissued_at":"2026-07-05T09:18:51.706380Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:18:51.706380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Long Context Compression with Activation Beacon","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ninglu Shao, Peitian Zhang, Qiwei Ye, Shitao Xiao, Zheng Liu, Zhicheng Dou","submitted_at":"2024-01-07T11:57:40Z","abstract_excerpt":"Long context compression is a critical research problem due to its significance in reducing the high computational and memory costs associated with LLMs. In this paper, we propose Activation Beacon, a plug-in module for transformer-based LLMs that targets effective, efficient, and flexible compression of long contexts. To achieve this, our method introduces the following technical designs. 1) We directly compress the activations (i.e. keys and values at every layer), rather than leveraging soft prompts to relay information (which constitute a major bottleneck to encapsulate the complex informa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.03462","kind":"arxiv","version":3},"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/2401.03462/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":"2401.03462","created_at":"2026-07-05T09:18:51.706441+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.03462v3","created_at":"2026-07-05T09:18:51.706441+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.03462","created_at":"2026-07-05T09:18:51.706441+00:00"},{"alias_kind":"pith_short_12","alias_value":"4FBCUMWK6YCR","created_at":"2026-07-05T09:18:51.706441+00:00"},{"alias_kind":"pith_short_16","alias_value":"4FBCUMWK6YCRT7SH","created_at":"2026-07-05T09:18:51.706441+00:00"},{"alias_kind":"pith_short_8","alias_value":"4FBCUMWK","created_at":"2026-07-05T09:18:51.706441+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"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":147,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14589","citing_title":"EndPrompt: Efficient Long-Context Extension via Terminal Anchoring","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24930","citing_title":"H$^{2}$MT: Semantic Hierarchy-Aware Hierarchical Memory Transformer","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28349","citing_title":"HMARS: A Hierarchical Multi-Agent Memory System for Long-Context Reasoning","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28713","citing_title":"Thinking as Compression: Your Reasoning Model is Secretly a Context Compressor","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2402.13753","citing_title":"LongRoPE: Extending LLM Context Window Beyond 2 Million Tokens","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14589","citing_title":"EndPrompt: Efficient Long-Context Extension via Terminal Anchoring","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20920","citing_title":"Simplified Sparse Attention via Gist Tokens","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2404.06654","citing_title":"RULER: What's the Real Context Size of Your Long-Context Language Models?","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T","json":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T.json","graph_json":"https://pith.science/api/pith-number/4FBCUMWK6YCRT7SHBFVRD2ZN5T/graph.json","events_json":"https://pith.science/api/pith-number/4FBCUMWK6YCRT7SHBFVRD2ZN5T/events.json","paper":"https://pith.science/paper/4FBCUMWK"},"agent_actions":{"view_html":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T","download_json":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T.json","view_paper":"https://pith.science/paper/4FBCUMWK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.03462&json=true","fetch_graph":"https://pith.science/api/pith-number/4FBCUMWK6YCRT7SHBFVRD2ZN5T/graph.json","fetch_events":"https://pith.science/api/pith-number/4FBCUMWK6YCRT7SHBFVRD2ZN5T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T/action/storage_attestation","attest_author":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T/action/author_attestation","sign_citation":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T/action/citation_signature","submit_replication":"https://pith.science/pith/4FBCUMWK6YCRT7SHBFVRD2ZN5T/action/replication_record"}},"created_at":"2026-07-05T09:18:51.706441+00:00","updated_at":"2026-07-05T09:18:51.706441+00:00"}