{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IGVG3EXBRV6REICKQITQFQH3IF","short_pith_number":"pith:IGVG3EXB","schema_version":"1.0","canonical_sha256":"41aa6d92e18d7d12204a822702c0fb41490d6330ac4a0982557e7828f3d4c313","source":{"kind":"arxiv","id":"2504.12526","version":1},"attestation_state":"computed","paper":{"title":"MOM: Memory-Efficient Offloaded Mini-Sequence Inference for Long Context Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Anima Anandkumar, Cheng Luo, Junyang Zhang, Tianyi Zhu","submitted_at":"2025-04-16T23:15:09Z","abstract_excerpt":"Long-context language models exhibit impressive performance but remain challenging to deploy due to high GPU memory demands during inference. We propose Memory-efficient Offloaded Mini-sequence Inference (MOM), a method that partitions critical layers into smaller \"mini-sequences\" and integrates seamlessly with KV cache offloading. Experiments on various Llama, Qwen, and Mistral models demonstrate that MOM reduces peak memory usage by over 50\\% on average. On Meta-Llama-3.2-8B, MOM extends the maximum context length from 155k to 455k tokens on a single A100 80GB GPU, while keeping outputs iden"},"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":"2504.12526","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-16T23:15:09Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"801660522fa7d163aa430f7f68592a7e4d8e77810402b0a62f2e32746dbf1de6","abstract_canon_sha256":"df57fb473746a54bc74dbde3e81d0127465394c86d0af2c8f27a5833187dd56a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:50:33.848923Z","signature_b64":"KD6h3l2sf00FYx9lVVysihWB4Eav+mKoi9S46afhS98b1+gpJhesZUt7mWMmsmv5mJMMAWNCidp9wwikb3QaAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"41aa6d92e18d7d12204a822702c0fb41490d6330ac4a0982557e7828f3d4c313","last_reissued_at":"2026-07-05T10:50:33.848433Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:50:33.848433Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MOM: Memory-Efficient Offloaded Mini-Sequence Inference for Long Context Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Anima Anandkumar, Cheng Luo, Junyang Zhang, Tianyi Zhu","submitted_at":"2025-04-16T23:15:09Z","abstract_excerpt":"Long-context language models exhibit impressive performance but remain challenging to deploy due to high GPU memory demands during inference. We propose Memory-efficient Offloaded Mini-sequence Inference (MOM), a method that partitions critical layers into smaller \"mini-sequences\" and integrates seamlessly with KV cache offloading. Experiments on various Llama, Qwen, and Mistral models demonstrate that MOM reduces peak memory usage by over 50\\% on average. On Meta-Llama-3.2-8B, MOM extends the maximum context length from 155k to 455k tokens on a single A100 80GB GPU, while keeping outputs iden"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.12526","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/2504.12526/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":"2504.12526","created_at":"2026-07-05T10:50:33.848500+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.12526v1","created_at":"2026-07-05T10:50:33.848500+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.12526","created_at":"2026-07-05T10:50:33.848500+00:00"},{"alias_kind":"pith_short_12","alias_value":"IGVG3EXBRV6R","created_at":"2026-07-05T10:50:33.848500+00:00"},{"alias_kind":"pith_short_16","alias_value":"IGVG3EXBRV6REICK","created_at":"2026-07-05T10:50:33.848500+00:00"},{"alias_kind":"pith_short_8","alias_value":"IGVG3EXB","created_at":"2026-07-05T10:50:33.848500+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/IGVG3EXBRV6REICKQITQFQH3IF","json":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF.json","graph_json":"https://pith.science/api/pith-number/IGVG3EXBRV6REICKQITQFQH3IF/graph.json","events_json":"https://pith.science/api/pith-number/IGVG3EXBRV6REICKQITQFQH3IF/events.json","paper":"https://pith.science/paper/IGVG3EXB"},"agent_actions":{"view_html":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF","download_json":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF.json","view_paper":"https://pith.science/paper/IGVG3EXB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.12526&json=true","fetch_graph":"https://pith.science/api/pith-number/IGVG3EXBRV6REICKQITQFQH3IF/graph.json","fetch_events":"https://pith.science/api/pith-number/IGVG3EXBRV6REICKQITQFQH3IF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF/action/storage_attestation","attest_author":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF/action/author_attestation","sign_citation":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF/action/citation_signature","submit_replication":"https://pith.science/pith/IGVG3EXBRV6REICKQITQFQH3IF/action/replication_record"}},"created_at":"2026-07-05T10:50:33.848500+00:00","updated_at":"2026-07-05T10:50:33.848500+00:00"}