{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:B4BJQ7JF4RUKZPKB2LDXRUNZ3J","short_pith_number":"pith:B4BJQ7JF","schema_version":"1.0","canonical_sha256":"0f02987d25e468acbd41d2c778d1b9da6f34abde72b3f4007412d57f5f5bfe95","source":{"kind":"arxiv","id":"2606.25091","version":1},"attestation_state":"computed","paper":{"title":"Speculation at a Distance: Where Edge-Cloud Speculative Decoding Actually Pays Off","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Bharath Irukulapati, Jaya Prakash Champati, Yuan Lyu","submitted_at":"2026-06-23T18:55:18Z","abstract_excerpt":"Speculative decoding (SD) accelerates LLM inference by $1.5$-$3$ times when the draft and target models are co-located. This has motivated a distributed variant (DSD) that places the draft model on an edge device while the target stays in the cloud. We show with closed-form inequalities that DSD's per-request latency benefit is limited under WAN edge-cloud communication. If the server can host both models, co-located SD has lower latency and communication than synchronous DSD, with the same per-output FLOPs and model-weight memory. Pipelining can make DSD competitive with co-located SD only in"},"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":"2606.25091","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2026-06-23T18:55:18Z","cross_cats_sorted":[],"title_canon_sha256":"2deaea6bbae1da12affb6ec8ee5d96cc891d3c9e973d80a5d87e2f690dd4bbab","abstract_canon_sha256":"9e6e54e80be9a6eca1e3f55aa9262fc017d351747bd515297d2ac51dc78a83f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-25T00:18:17.565224Z","signature_b64":"7/60BS9iBpMKIwZubCtjgu114FmDyM39g/HDBYXQDRiv1q10eDxyHOdIUDnhfhcQW2A9fMojAwg0QPjkq2RdDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0f02987d25e468acbd41d2c778d1b9da6f34abde72b3f4007412d57f5f5bfe95","last_reissued_at":"2026-06-25T00:18:17.564858Z","signature_status":"signed_v1","first_computed_at":"2026-06-25T00:18:17.564858Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Speculation at a Distance: Where Edge-Cloud Speculative Decoding Actually Pays Off","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Bharath Irukulapati, Jaya Prakash Champati, Yuan Lyu","submitted_at":"2026-06-23T18:55:18Z","abstract_excerpt":"Speculative decoding (SD) accelerates LLM inference by $1.5$-$3$ times when the draft and target models are co-located. This has motivated a distributed variant (DSD) that places the draft model on an edge device while the target stays in the cloud. We show with closed-form inequalities that DSD's per-request latency benefit is limited under WAN edge-cloud communication. If the server can host both models, co-located SD has lower latency and communication than synchronous DSD, with the same per-output FLOPs and model-weight memory. Pipelining can make DSD competitive with co-located SD only in"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.25091","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/2606.25091/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":"2606.25091","created_at":"2026-06-25T00:18:17.564925+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.25091v1","created_at":"2026-06-25T00:18:17.564925+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.25091","created_at":"2026-06-25T00:18:17.564925+00:00"},{"alias_kind":"pith_short_12","alias_value":"B4BJQ7JF4RUK","created_at":"2026-06-25T00:18:17.564925+00:00"},{"alias_kind":"pith_short_16","alias_value":"B4BJQ7JF4RUKZPKB","created_at":"2026-06-25T00:18:17.564925+00:00"},{"alias_kind":"pith_short_8","alias_value":"B4BJQ7JF","created_at":"2026-06-25T00:18:17.564925+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/B4BJQ7JF4RUKZPKB2LDXRUNZ3J","json":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J.json","graph_json":"https://pith.science/api/pith-number/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/graph.json","events_json":"https://pith.science/api/pith-number/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/events.json","paper":"https://pith.science/paper/B4BJQ7JF"},"agent_actions":{"view_html":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J","download_json":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J.json","view_paper":"https://pith.science/paper/B4BJQ7JF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.25091&json=true","fetch_graph":"https://pith.science/api/pith-number/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/graph.json","fetch_events":"https://pith.science/api/pith-number/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/action/storage_attestation","attest_author":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/action/author_attestation","sign_citation":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/action/citation_signature","submit_replication":"https://pith.science/pith/B4BJQ7JF4RUKZPKB2LDXRUNZ3J/action/replication_record"}},"created_at":"2026-06-25T00:18:17.564925+00:00","updated_at":"2026-06-25T00:18:17.564925+00:00"}