{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:CZD33ZP2TJRAQSRSPZHK35Q5QG","short_pith_number":"pith:CZD33ZP2","schema_version":"1.0","canonical_sha256":"1647bde5fa9a62084a327e4eadf61d81826c1005a43e5a677f4b297c6815f741","source":{"kind":"arxiv","id":"2403.20041","version":3},"attestation_state":"computed","paper":{"title":"Transformer-Lite: High-efficiency Deployment of Large Language Models on Mobile Phone GPUs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jie Lu, Luchang Li, Lunxi Yuan, Qin Xie, Rui Wang, Sheng Qian","submitted_at":"2024-03-29T08:26:53Z","abstract_excerpt":"The Large Language Model (LLM) is widely employed for tasks such as intelligent assistants, text summarization, translation, and multi-modality on mobile phones. However, the current methods for on-device LLM deployment maintain slow inference speed, which causes poor user experience. To facilitate high-efficiency LLM deployment on device GPUs, we propose four optimization techniques: (a) a symbolic expression-based approach to support dynamic shape model inference; (b) operator optimizations and execution priority setting to enhance inference speed and reduce phone lagging; (c) an FP4 quantiz"},"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":"2403.20041","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-29T08:26:53Z","cross_cats_sorted":[],"title_canon_sha256":"21fc1579d9debc444d64e1ed4b7943ad923bbd6db07090b06db83d9c994620ef","abstract_canon_sha256":"5d22ed8a55fc8bcb1a0ed7835b1074f87284cd6ded98969d8ea9c364dc711838"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:40:15.754701Z","signature_b64":"fSUxYCXUbPxFgLct6WYzU64I9aCT3FkdeBjWQcck0FrigNAuVWB8oS2GHxqXhFdZcroOmlX707U3V65X31sUBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1647bde5fa9a62084a327e4eadf61d81826c1005a43e5a677f4b297c6815f741","last_reissued_at":"2026-07-05T08:40:15.754195Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:40:15.754195Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Transformer-Lite: High-efficiency Deployment of Large Language Models on Mobile Phone GPUs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jie Lu, Luchang Li, Lunxi Yuan, Qin Xie, Rui Wang, Sheng Qian","submitted_at":"2024-03-29T08:26:53Z","abstract_excerpt":"The Large Language Model (LLM) is widely employed for tasks such as intelligent assistants, text summarization, translation, and multi-modality on mobile phones. However, the current methods for on-device LLM deployment maintain slow inference speed, which causes poor user experience. To facilitate high-efficiency LLM deployment on device GPUs, we propose four optimization techniques: (a) a symbolic expression-based approach to support dynamic shape model inference; (b) operator optimizations and execution priority setting to enhance inference speed and reduce phone lagging; (c) an FP4 quantiz"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.20041","kind":"arxiv","version":3},"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/2403.20041/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":"2403.20041","created_at":"2026-07-05T08:40:15.754255+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.20041v3","created_at":"2026-07-05T08:40:15.754255+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.20041","created_at":"2026-07-05T08:40:15.754255+00:00"},{"alias_kind":"pith_short_12","alias_value":"CZD33ZP2TJRA","created_at":"2026-07-05T08:40:15.754255+00:00"},{"alias_kind":"pith_short_16","alias_value":"CZD33ZP2TJRAQSRS","created_at":"2026-07-05T08:40:15.754255+00:00"},{"alias_kind":"pith_short_8","alias_value":"CZD33ZP2","created_at":"2026-07-05T08:40:15.754255+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.09083","citing_title":"EdgeFlow: Fast Cold Starts for LLMs on Mobile Devices","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG","json":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG.json","graph_json":"https://pith.science/api/pith-number/CZD33ZP2TJRAQSRSPZHK35Q5QG/graph.json","events_json":"https://pith.science/api/pith-number/CZD33ZP2TJRAQSRSPZHK35Q5QG/events.json","paper":"https://pith.science/paper/CZD33ZP2"},"agent_actions":{"view_html":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG","download_json":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG.json","view_paper":"https://pith.science/paper/CZD33ZP2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.20041&json=true","fetch_graph":"https://pith.science/api/pith-number/CZD33ZP2TJRAQSRSPZHK35Q5QG/graph.json","fetch_events":"https://pith.science/api/pith-number/CZD33ZP2TJRAQSRSPZHK35Q5QG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG/action/storage_attestation","attest_author":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG/action/author_attestation","sign_citation":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG/action/citation_signature","submit_replication":"https://pith.science/pith/CZD33ZP2TJRAQSRSPZHK35Q5QG/action/replication_record"}},"created_at":"2026-07-05T08:40:15.754255+00:00","updated_at":"2026-07-05T08:40:15.754255+00:00"}