{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:Y44ARQ3A7R37MA3UJEILH3MPVI","short_pith_number":"pith:Y44ARQ3A","schema_version":"1.0","canonical_sha256":"c73808c360fc77f603744910b3ed8faa0396346407aff1d37b969020be0b9b9d","source":{"kind":"arxiv","id":"2501.06589","version":5},"attestation_state":"computed","paper":{"title":"Ladder-residual: parallelism-aware architecture for accelerating large model inference with communication overlapping","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.DC"],"primary_cat":"cs.LG","authors_text":"Ben Athiwaratkun, Jonathan Ragan-Kelley, Jue Wang, Mayank Mishra, Muru Zhang, Shuaiwen Leon Song, Tri Dao, William Brandon, Yoon Kim, Zhongzhu Zhou","submitted_at":"2025-01-11T17:06:30Z","abstract_excerpt":"Large language model inference is both memory-intensive and time-consuming, often requiring distributed algorithms to efficiently scale. Various model parallelism strategies are used in multi-gpu training and inference to partition computation across multiple devices, reducing memory load and computation time. However, using model parallelism necessitates communication of information between GPUs, which has been a major bottleneck and limits the gains obtained by scaling up the number of devices. We introduce Ladder Residual, a simple architectural modification applicable to all residual-based"},"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":"2501.06589","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-11T17:06:30Z","cross_cats_sorted":["cs.CL","cs.DC"],"title_canon_sha256":"5c8632ad6e4918920dd9f0671be888426f33eb3e25a3d96ee4a0a4c10b024aea","abstract_canon_sha256":"e99aa1cd8c245f0ea8b4888e3476b82b9ce925ca8218363fbb03e6441197ee21"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:12.550654Z","signature_b64":"pfn8Wh+kJAxaKyuONRcTGmQm2Kq91Y9RQKrAR4qu/kVTloJVviUogk62Zl0xe90YxV+z5kZZALgmw6IKmUrnAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c73808c360fc77f603744910b3ed8faa0396346407aff1d37b969020be0b9b9d","last_reissued_at":"2026-07-05T11:24:12.550086Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:12.550086Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Ladder-residual: parallelism-aware architecture for accelerating large model inference with communication overlapping","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.DC"],"primary_cat":"cs.LG","authors_text":"Ben Athiwaratkun, Jonathan Ragan-Kelley, Jue Wang, Mayank Mishra, Muru Zhang, Shuaiwen Leon Song, Tri Dao, William Brandon, Yoon Kim, Zhongzhu Zhou","submitted_at":"2025-01-11T17:06:30Z","abstract_excerpt":"Large language model inference is both memory-intensive and time-consuming, often requiring distributed algorithms to efficiently scale. Various model parallelism strategies are used in multi-gpu training and inference to partition computation across multiple devices, reducing memory load and computation time. However, using model parallelism necessitates communication of information between GPUs, which has been a major bottleneck and limits the gains obtained by scaling up the number of devices. We introduce Ladder Residual, a simple architectural modification applicable to all residual-based"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.06589","kind":"arxiv","version":5},"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/2501.06589/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":"2501.06589","created_at":"2026-07-05T11:24:12.550151+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.06589v5","created_at":"2026-07-05T11:24:12.550151+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.06589","created_at":"2026-07-05T11:24:12.550151+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y44ARQ3A7R37","created_at":"2026-07-05T11:24:12.550151+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y44ARQ3A7R37MA3U","created_at":"2026-07-05T11:24:12.550151+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y44ARQ3A","created_at":"2026-07-05T11:24:12.550151+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2511.09557","citing_title":"Understanding and Improving Communication Performance in Multi-node LLM Inference","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2603.14360","citing_title":"M$^2$RNN: Non-Linear RNNs with Matrix-Valued States for Scalable Language Modeling","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI","json":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI.json","graph_json":"https://pith.science/api/pith-number/Y44ARQ3A7R37MA3UJEILH3MPVI/graph.json","events_json":"https://pith.science/api/pith-number/Y44ARQ3A7R37MA3UJEILH3MPVI/events.json","paper":"https://pith.science/paper/Y44ARQ3A"},"agent_actions":{"view_html":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI","download_json":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI.json","view_paper":"https://pith.science/paper/Y44ARQ3A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.06589&json=true","fetch_graph":"https://pith.science/api/pith-number/Y44ARQ3A7R37MA3UJEILH3MPVI/graph.json","fetch_events":"https://pith.science/api/pith-number/Y44ARQ3A7R37MA3UJEILH3MPVI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI/action/storage_attestation","attest_author":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI/action/author_attestation","sign_citation":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI/action/citation_signature","submit_replication":"https://pith.science/pith/Y44ARQ3A7R37MA3UJEILH3MPVI/action/replication_record"}},"created_at":"2026-07-05T11:24:12.550151+00:00","updated_at":"2026-07-05T11:24:12.550151+00:00"}