{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:4MJCW2CW2IJI4KFD2N5O7KXTYZ","short_pith_number":"pith:4MJCW2CW","schema_version":"1.0","canonical_sha256":"e3122b6856d2128e28a3d37aefaaf3c64a64391adaa6e2328e2fb5b0f2bf1db0","source":{"kind":"arxiv","id":"2606.23521","version":1},"attestation_state":"computed","paper":{"title":"Concordia: JIT-Compiled Persistent-Kernel Checkpointing for Fault-Tolerant LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Andi Quinn, Chen Qian, Rain Jiang, Xiangyu Gao, Xiaoning Ding, Yichen Wang, Yiwei Yang, Yuhang Gan, Yuyi Li","submitted_at":"2026-06-22T16:06:11Z","abstract_excerpt":"Long-running LLM agents keep valuable state resident on GPUs: KV caches, request schedulers, communication state, and sometimes online adapters. Losing this state after a GPU or communicator failure can discard minutes to hours of work, yet existing recovery mechanisms either restart the whole serving stack or require application-specific checkpoint logic inside every attention and runtime component. This paper argues that fault tolerance for such workloads needs a GPU-resident execution context: checkpoint hooks must run at device synchronization points, observe binary kernels that frameworks"},"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":"2606.23521","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-06-22T16:06:11Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"7b05ba7592ea395f1b5eb70ad383ffebbc8f15346a3b9e6ccd3dc1467d881bf1","abstract_canon_sha256":"1179ebf880c47608e9030c81c4580d77e46235b21d9edd3e8031deb58911728a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:14:22.157329Z","signature_b64":"YZzNPLrhFSAmdmf8ZyStoFh4dhMdws6NwNdP5EVDwmjG/mbHa6aKBJSQeyaSUGiEdRCmRqnxQZ1+PHK5iEN0Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e3122b6856d2128e28a3d37aefaaf3c64a64391adaa6e2328e2fb5b0f2bf1db0","last_reissued_at":"2026-06-23T03:14:22.156953Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:14:22.156953Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Concordia: JIT-Compiled Persistent-Kernel Checkpointing for Fault-Tolerant LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Andi Quinn, Chen Qian, Rain Jiang, Xiangyu Gao, Xiaoning Ding, Yichen Wang, Yiwei Yang, Yuhang Gan, Yuyi Li","submitted_at":"2026-06-22T16:06:11Z","abstract_excerpt":"Long-running LLM agents keep valuable state resident on GPUs: KV caches, request schedulers, communication state, and sometimes online adapters. Losing this state after a GPU or communicator failure can discard minutes to hours of work, yet existing recovery mechanisms either restart the whole serving stack or require application-specific checkpoint logic inside every attention and runtime component. This paper argues that fault tolerance for such workloads needs a GPU-resident execution context: checkpoint hooks must run at device synchronization points, observe binary kernels that frameworks"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.23521","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/2606.23521/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":"2606.23521","created_at":"2026-06-23T03:14:22.157010+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.23521v1","created_at":"2026-06-23T03:14:22.157010+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.23521","created_at":"2026-06-23T03:14:22.157010+00:00"},{"alias_kind":"pith_short_12","alias_value":"4MJCW2CW2IJI","created_at":"2026-06-23T03:14:22.157010+00:00"},{"alias_kind":"pith_short_16","alias_value":"4MJCW2CW2IJI4KFD","created_at":"2026-06-23T03:14:22.157010+00:00"},{"alias_kind":"pith_short_8","alias_value":"4MJCW2CW","created_at":"2026-06-23T03:14:22.157010+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/4MJCW2CW2IJI4KFD2N5O7KXTYZ","json":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ.json","graph_json":"https://pith.science/api/pith-number/4MJCW2CW2IJI4KFD2N5O7KXTYZ/graph.json","events_json":"https://pith.science/api/pith-number/4MJCW2CW2IJI4KFD2N5O7KXTYZ/events.json","paper":"https://pith.science/paper/4MJCW2CW"},"agent_actions":{"view_html":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ","download_json":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ.json","view_paper":"https://pith.science/paper/4MJCW2CW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.23521&json=true","fetch_graph":"https://pith.science/api/pith-number/4MJCW2CW2IJI4KFD2N5O7KXTYZ/graph.json","fetch_events":"https://pith.science/api/pith-number/4MJCW2CW2IJI4KFD2N5O7KXTYZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ/action/storage_attestation","attest_author":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ/action/author_attestation","sign_citation":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ/action/citation_signature","submit_replication":"https://pith.science/pith/4MJCW2CW2IJI4KFD2N5O7KXTYZ/action/replication_record"}},"created_at":"2026-06-23T03:14:22.157010+00:00","updated_at":"2026-06-23T03:14:22.157010+00:00"}