{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JC54UBGXH2BLRBCJMKBAHTVZYJ","short_pith_number":"pith:JC54UBGX","schema_version":"1.0","canonical_sha256":"48bbca04d73e82b88449628203ceb9c24ab7d172a45998fda617ff2c189edb33","source":{"kind":"arxiv","id":"2505.00232","version":1},"attestation_state":"computed","paper":{"title":"Scaling On-Device GPU Inference for Large Generative Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andrei Kulik, Ekaterina Ignasheva, Grant Jensen, Jiuqiang Tang, Juhyun Lee, Lin Chen, Matthias Grundmann, Raman Sarokin","submitted_at":"2025-05-01T00:44:13Z","abstract_excerpt":"Driven by the advancements in generative AI, large machine learning models have revolutionized domains such as image processing, audio synthesis, and speech recognition. While server-based deployments remain the locus of peak performance, the imperative for on-device inference, necessitated by privacy and efficiency considerations, persists. Recognizing GPUs as the on-device ML accelerator with the widest reach, we present ML Drift--an optimized framework that extends the capabilities of state-of-the-art GPU-accelerated inference engines. ML Drift enables on-device execution of generative AI w"},"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":"2505.00232","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-01T00:44:13Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dcff3147e74d0c8f6110635f150676c294091222e8f7dcea32cc1e8714536513","abstract_canon_sha256":"f80919641f1bae3a45bfc3bc3a25194c4510dcd84baa5d07adaacf3dadce128a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:56:45.114171Z","signature_b64":"2LGHauQ63WhyCSqOYzf8QA3xJe0bj9pPrabRlYj+5H9KyLZmghukvc/5o88ZPi3mlY4irhi55qz+P12Oub6hCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"48bbca04d73e82b88449628203ceb9c24ab7d172a45998fda617ff2c189edb33","last_reissued_at":"2026-07-05T10:56:45.113710Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:56:45.113710Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling On-Device GPU Inference for Large Generative Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Andrei Kulik, Ekaterina Ignasheva, Grant Jensen, Jiuqiang Tang, Juhyun Lee, Lin Chen, Matthias Grundmann, Raman Sarokin","submitted_at":"2025-05-01T00:44:13Z","abstract_excerpt":"Driven by the advancements in generative AI, large machine learning models have revolutionized domains such as image processing, audio synthesis, and speech recognition. While server-based deployments remain the locus of peak performance, the imperative for on-device inference, necessitated by privacy and efficiency considerations, persists. Recognizing GPUs as the on-device ML accelerator with the widest reach, we present ML Drift--an optimized framework that extends the capabilities of state-of-the-art GPU-accelerated inference engines. ML Drift enables on-device execution of generative AI w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.00232","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/2505.00232/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":"2505.00232","created_at":"2026-07-05T10:56:45.113763+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.00232v1","created_at":"2026-07-05T10:56:45.113763+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.00232","created_at":"2026-07-05T10:56:45.113763+00:00"},{"alias_kind":"pith_short_12","alias_value":"JC54UBGXH2BL","created_at":"2026-07-05T10:56:45.113763+00:00"},{"alias_kind":"pith_short_16","alias_value":"JC54UBGXH2BLRBCJ","created_at":"2026-07-05T10:56:45.113763+00:00"},{"alias_kind":"pith_short_8","alias_value":"JC54UBGX","created_at":"2026-07-05T10:56:45.113763+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.11417","citing_title":"EdgeWisePersona: A Dataset for On-Device User Profiling from Natural Language Interactions","ref_index":32,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ","json":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ.json","graph_json":"https://pith.science/api/pith-number/JC54UBGXH2BLRBCJMKBAHTVZYJ/graph.json","events_json":"https://pith.science/api/pith-number/JC54UBGXH2BLRBCJMKBAHTVZYJ/events.json","paper":"https://pith.science/paper/JC54UBGX"},"agent_actions":{"view_html":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ","download_json":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ.json","view_paper":"https://pith.science/paper/JC54UBGX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.00232&json=true","fetch_graph":"https://pith.science/api/pith-number/JC54UBGXH2BLRBCJMKBAHTVZYJ/graph.json","fetch_events":"https://pith.science/api/pith-number/JC54UBGXH2BLRBCJMKBAHTVZYJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ/action/storage_attestation","attest_author":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ/action/author_attestation","sign_citation":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ/action/citation_signature","submit_replication":"https://pith.science/pith/JC54UBGXH2BLRBCJMKBAHTVZYJ/action/replication_record"}},"created_at":"2026-07-05T10:56:45.113763+00:00","updated_at":"2026-07-05T10:56:45.113763+00:00"}