{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FQDUY6JE47SWPDZMHZ5QWLW4UT","short_pith_number":"pith:FQDUY6JE","schema_version":"1.0","canonical_sha256":"2c074c7924e7e5678f2c3e7b0b2edca4c9ce82eeda03dc463ef6ff1340ea3d30","source":{"kind":"arxiv","id":"2608.05303","version":1},"attestation_state":"computed","paper":{"title":"EdgeXpert: An Edge Device for Memory-Efficient LLM Inference with Mixture-of-Experts and Speculative Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AR","authors_text":"Hoi-Jun Yoo, Hyunwoo Seo, Sangwoo Ha, Youngjin Moon, Yurim Jo","submitted_at":"2026-08-05T18:03:47Z","abstract_excerpt":"On-device deployment of Large Language Models (LLMs) has become essential for personalized edge applications. A primary bottleneck is external memory access (EMA) in feed-forward network (FFN) layers. Speculative decoding and mixture-of-experts (MoE) are promising solutions. Speculative decoding reduces the number of decoding stages by generating multiple tokens per stage, and MoE minimizes per-stage cost through sparse expert activation. However, there is an incompatibility when combining these two techniques. We propose EdgeXpert, a software-hardware co-designed LLM accelerator that resolves"},"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":"2608.05303","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2026-08-05T18:03:47Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"2f33173199ec91b6b3eb5daebf8791df0f201a2f47c0cbd8436fcc0ee77038fc","abstract_canon_sha256":"f3165b6af5fc8b951e021a7a85291eaf79bca9155e169494f52f78b884977fc7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-07T00:46:58.453204Z","signature_b64":"gb0UVr8JZiv0XhPjJVXfvHx86RKHQOurEE76DHOjA5O8hhh8MZolBlcIiFOdA1ncTzXgu7uYhFOY8E1YIIfODg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c074c7924e7e5678f2c3e7b0b2edca4c9ce82eeda03dc463ef6ff1340ea3d30","last_reissued_at":"2026-08-07T00:46:58.451417Z","signature_status":"signed_v1","first_computed_at":"2026-08-07T00:46:58.451417Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"EdgeXpert: An Edge Device for Memory-Efficient LLM Inference with Mixture-of-Experts and Speculative Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AR","authors_text":"Hoi-Jun Yoo, Hyunwoo Seo, Sangwoo Ha, Youngjin Moon, Yurim Jo","submitted_at":"2026-08-05T18:03:47Z","abstract_excerpt":"On-device deployment of Large Language Models (LLMs) has become essential for personalized edge applications. A primary bottleneck is external memory access (EMA) in feed-forward network (FFN) layers. Speculative decoding and mixture-of-experts (MoE) are promising solutions. Speculative decoding reduces the number of decoding stages by generating multiple tokens per stage, and MoE minimizes per-stage cost through sparse expert activation. However, there is an incompatibility when combining these two techniques. We propose EdgeXpert, a software-hardware co-designed LLM accelerator that resolves"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.05303","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/2608.05303/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":"2608.05303","created_at":"2026-08-07T00:46:58.452825+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.05303v1","created_at":"2026-08-07T00:46:58.452825+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.05303","created_at":"2026-08-07T00:46:58.452825+00:00"},{"alias_kind":"pith_short_12","alias_value":"FQDUY6JE47SW","created_at":"2026-08-07T00:46:58.452825+00:00"},{"alias_kind":"pith_short_16","alias_value":"FQDUY6JE47SWPDZM","created_at":"2026-08-07T00:46:58.452825+00:00"},{"alias_kind":"pith_short_8","alias_value":"FQDUY6JE","created_at":"2026-08-07T00:46:58.452825+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/FQDUY6JE47SWPDZMHZ5QWLW4UT","json":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT.json","graph_json":"https://pith.science/api/pith-number/FQDUY6JE47SWPDZMHZ5QWLW4UT/graph.json","events_json":"https://pith.science/api/pith-number/FQDUY6JE47SWPDZMHZ5QWLW4UT/events.json","paper":"https://pith.science/paper/FQDUY6JE"},"agent_actions":{"view_html":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT","download_json":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT.json","view_paper":"https://pith.science/paper/FQDUY6JE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.05303&json=true","fetch_graph":"https://pith.science/api/pith-number/FQDUY6JE47SWPDZMHZ5QWLW4UT/graph.json","fetch_events":"https://pith.science/api/pith-number/FQDUY6JE47SWPDZMHZ5QWLW4UT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT/action/storage_attestation","attest_author":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT/action/author_attestation","sign_citation":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT/action/citation_signature","submit_replication":"https://pith.science/pith/FQDUY6JE47SWPDZMHZ5QWLW4UT/action/replication_record"}},"created_at":"2026-08-07T00:46:58.452825+00:00","updated_at":"2026-08-07T00:46:58.452825+00:00"}