{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OQSXWL2NNIP5X7TVA7D42KC4LQ","short_pith_number":"pith:OQSXWL2N","schema_version":"1.0","canonical_sha256":"74257b2f4d6a1fdbfe7507c7cd285c5c1c3044189cb8037145aee121360915a5","source":{"kind":"arxiv","id":"2604.00491","version":2},"attestation_state":"computed","paper":{"title":"Executing as You Generate: Hiding Execution Latency in LLM Code Interpreters","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SE"],"primary_cat":"cs.PL","authors_text":"Chengran Yang, David Lo, Li Li, Mingyi Zhou, Zhensu Sun, Zhi Chen, Zhihao Lin","submitted_at":"2026-04-01T05:17:53Z","abstract_excerpt":"Current LLM systems are increasingly equipped with a code interpreter that executes generated code to obtain results. This works serially: the model first generates the complete code, then an interpreter executes it. This sequential workflow leaves the executor idle during generation and the generator idle during execution, resulting in unnecessary end-to-end latency. Our key observation is that an LLM, unlike a human developer, emits code tokens left to right and does not backtrack over what it has already written. This makes it possible to start executing a piece of code while later tokens a"},"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":"2604.00491","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.PL","submitted_at":"2026-04-01T05:17:53Z","cross_cats_sorted":["cs.AI","cs.SE"],"title_canon_sha256":"74b49dfcb25f3ee5d16ed45b346eeb73c17ee3e0f828051438a351c7caf93cf5","abstract_canon_sha256":"5c9db387e7ecae8954714a78c63d9e0c65ff7c09cd31f0ddd1f96b1c7e682e55"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:13:56.856505Z","signature_b64":"YwT4OhWTcpj7abn0swEbMR5na8NPZIUCjjQdIn1h/JGY9f7nZvelmo6iRtp95GKZhjsgbHuEx4lPITFYgI3FDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"74257b2f4d6a1fdbfe7507c7cd285c5c1c3044189cb8037145aee121360915a5","last_reissued_at":"2026-06-23T03:13:56.856076Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:13:56.856076Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Executing as You Generate: Hiding Execution Latency in LLM Code Interpreters","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SE"],"primary_cat":"cs.PL","authors_text":"Chengran Yang, David Lo, Li Li, Mingyi Zhou, Zhensu Sun, Zhi Chen, Zhihao Lin","submitted_at":"2026-04-01T05:17:53Z","abstract_excerpt":"Current LLM systems are increasingly equipped with a code interpreter that executes generated code to obtain results. This works serially: the model first generates the complete code, then an interpreter executes it. This sequential workflow leaves the executor idle during generation and the generator idle during execution, resulting in unnecessary end-to-end latency. Our key observation is that an LLM, unlike a human developer, emits code tokens left to right and does not backtrack over what it has already written. This makes it possible to start executing a piece of code while later tokens a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.00491","kind":"arxiv","version":2},"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/2604.00491/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":"2604.00491","created_at":"2026-06-23T03:13:56.856137+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.00491v2","created_at":"2026-06-23T03:13:56.856137+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.00491","created_at":"2026-06-23T03:13:56.856137+00:00"},{"alias_kind":"pith_short_12","alias_value":"OQSXWL2NNIP5","created_at":"2026-06-23T03:13:56.856137+00:00"},{"alias_kind":"pith_short_16","alias_value":"OQSXWL2NNIP5X7TV","created_at":"2026-06-23T03:13:56.856137+00:00"},{"alias_kind":"pith_short_8","alias_value":"OQSXWL2N","created_at":"2026-06-23T03:13:56.856137+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.01211","citing_title":"Are Performance-Optimization Benchmarks Reliably Measuring Coding Agents?","ref_index":36,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ","json":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ.json","graph_json":"https://pith.science/api/pith-number/OQSXWL2NNIP5X7TVA7D42KC4LQ/graph.json","events_json":"https://pith.science/api/pith-number/OQSXWL2NNIP5X7TVA7D42KC4LQ/events.json","paper":"https://pith.science/paper/OQSXWL2N"},"agent_actions":{"view_html":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ","download_json":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ.json","view_paper":"https://pith.science/paper/OQSXWL2N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.00491&json=true","fetch_graph":"https://pith.science/api/pith-number/OQSXWL2NNIP5X7TVA7D42KC4LQ/graph.json","fetch_events":"https://pith.science/api/pith-number/OQSXWL2NNIP5X7TVA7D42KC4LQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ/action/storage_attestation","attest_author":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ/action/author_attestation","sign_citation":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ/action/citation_signature","submit_replication":"https://pith.science/pith/OQSXWL2NNIP5X7TVA7D42KC4LQ/action/replication_record"}},"created_at":"2026-06-23T03:13:56.856137+00:00","updated_at":"2026-06-23T03:13:56.856137+00:00"}