{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:74YUI3NHKFKQT5TSE46Q6SZ36V","short_pith_number":"pith:74YUI3NH","schema_version":"1.0","canonical_sha256":"ff31446da7515509f672273d0f4b3bf57b93d276573d857d439b48be8a5a7126","source":{"kind":"arxiv","id":"2310.10046","version":3},"attestation_state":"computed","paper":{"title":"TRANSOM: An Efficient Fault-Tolerant System for Training LLMs","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Baodong Wu, Kangyu Li, Lei Xia, Qingping Li, Shigang Li, Tieyao Xiang, Xu Chen, Yongqiang Guo, Yuheng Chen","submitted_at":"2023-10-16T04:06:52Z","abstract_excerpt":"Large language models (LLMs) with hundreds of billions or trillions of parameters, represented by chatGPT, have achieved profound impact on various fields. However, training LLMs with super-large-scale parameters requires large high-performance GPU clusters and long training periods lasting for months. Due to the inevitable hardware and software failures in large-scale clusters, maintaining uninterrupted and long-duration training is extremely challenging. As a result, A substantial amount of training time is devoted to task checkpoint saving and loading, task rescheduling and restart, and tas"},"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":"2310.10046","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.DC","submitted_at":"2023-10-16T04:06:52Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"003881551436091d6e0e5e57164795661a375d0387e5d271f8df2c907111b1a0","abstract_canon_sha256":"3da50343ee4ab8bf03a689b3e3682f7e8e5a187f7b6bcc133ac1ebdadc4aad41"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:02:06.438292Z","signature_b64":"yfSuRDYo1yUF54i8PV1N4gNtcDQ09s1h6dQtBb/wwKW118aI9h/BUkUpaSvnPf3IcU75jJfXETv1tZGJ1WJlAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff31446da7515509f672273d0f4b3bf57b93d276573d857d439b48be8a5a7126","last_reissued_at":"2026-07-05T07:02:06.437780Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:02:06.437780Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TRANSOM: An Efficient Fault-Tolerant System for Training LLMs","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Baodong Wu, Kangyu Li, Lei Xia, Qingping Li, Shigang Li, Tieyao Xiang, Xu Chen, Yongqiang Guo, Yuheng Chen","submitted_at":"2023-10-16T04:06:52Z","abstract_excerpt":"Large language models (LLMs) with hundreds of billions or trillions of parameters, represented by chatGPT, have achieved profound impact on various fields. However, training LLMs with super-large-scale parameters requires large high-performance GPU clusters and long training periods lasting for months. Due to the inevitable hardware and software failures in large-scale clusters, maintaining uninterrupted and long-duration training is extremely challenging. As a result, A substantial amount of training time is devoted to task checkpoint saving and loading, task rescheduling and restart, and tas"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.10046","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/2310.10046/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":"2310.10046","created_at":"2026-07-05T07:02:06.437847+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.10046v3","created_at":"2026-07-05T07:02:06.437847+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.10046","created_at":"2026-07-05T07:02:06.437847+00:00"},{"alias_kind":"pith_short_12","alias_value":"74YUI3NHKFKQ","created_at":"2026-07-05T07:02:06.437847+00:00"},{"alias_kind":"pith_short_16","alias_value":"74YUI3NHKFKQT5TS","created_at":"2026-07-05T07:02:06.437847+00:00"},{"alias_kind":"pith_short_8","alias_value":"74YUI3NH","created_at":"2026-07-05T07:02:06.437847+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01646","citing_title":"PHOENIX: Resilient LLM Training with Hot-Swapping via Zero-Overhead Checkpoint","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09370","citing_title":"From Detection to Recovery: Operational Analysis on LLM Pre-training with 504 GPUs","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17821","citing_title":"TierCheck: Tiered Checkpointing for Fault Tolerance in Large Language Model Training","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09370","citing_title":"From Detection to Recovery: Operational Analysis on LLM Pre-training with 504 GPUs","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04431","citing_title":"Towards Robust LLM Post-Training: Automatic Failure Management for Reinforcement Fine-Tuning","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V","json":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V.json","graph_json":"https://pith.science/api/pith-number/74YUI3NHKFKQT5TSE46Q6SZ36V/graph.json","events_json":"https://pith.science/api/pith-number/74YUI3NHKFKQT5TSE46Q6SZ36V/events.json","paper":"https://pith.science/paper/74YUI3NH"},"agent_actions":{"view_html":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V","download_json":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V.json","view_paper":"https://pith.science/paper/74YUI3NH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.10046&json=true","fetch_graph":"https://pith.science/api/pith-number/74YUI3NHKFKQT5TSE46Q6SZ36V/graph.json","fetch_events":"https://pith.science/api/pith-number/74YUI3NHKFKQT5TSE46Q6SZ36V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V/action/storage_attestation","attest_author":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V/action/author_attestation","sign_citation":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V/action/citation_signature","submit_replication":"https://pith.science/pith/74YUI3NHKFKQT5TSE46Q6SZ36V/action/replication_record"}},"created_at":"2026-07-05T07:02:06.437847+00:00","updated_at":"2026-07-05T07:02:06.437847+00:00"}