{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YXYZUDTEV5EOPMBC6RNLUNBRRP","short_pith_number":"pith:YXYZUDTE","schema_version":"1.0","canonical_sha256":"c5f19a0e64af48e7b022f45aba34318bcc8aad9e983bf815c25def34ad5b71cc","source":{"kind":"arxiv","id":"2410.06511","version":3},"attestation_state":"computed","paper":{"title":"TorchTitan: One-stop PyTorch native solution for production ready LLM pre-training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DC","cs.LG"],"primary_cat":"cs.CL","authors_text":"Andrew Gu, Chien-Chin Huang, Gokul Nadathur, Howard Huang, Iris Zhang, Junjie Wang, Less Wright, Sanket Purandare, Stratos Idreos, Tianyu Liu, Wanchao Liang, Wei Feng, Will Constable","submitted_at":"2024-10-09T03:26:11Z","abstract_excerpt":"The development of large language models (LLMs) has been instrumental in advancing state-of-the-art natural language processing applications. Training LLMs with billions of parameters and trillions of tokens require sophisticated distributed systems that enable composing and comparing several state-of-the-art techniques in order to efficiently scale across thousands of accelerators. However, existing solutions are complex, scattered across multiple libraries/repositories, lack interoperability, and are cumbersome to maintain. Thus, curating and empirically comparing training recipes require no"},"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":"2410.06511","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-09T03:26:11Z","cross_cats_sorted":["cs.AI","cs.DC","cs.LG"],"title_canon_sha256":"b8d0e69d465ead53afb70b976a74d7b8c4f19b4f26ed33b23bb0effc9f31e855","abstract_canon_sha256":"9612e3dee456133b924e3fd1d3c3d2638b31be5fdc3bfc88fca253b0c4cb2830"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:36.247519Z","signature_b64":"c999k3Q3LR2ClgFXW8JEXHQAkycCy8COm4+KZSRanzkki8kp3CXMrq7LmEtpjknuFy6Tb+0scWnsAQZFyzypAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5f19a0e64af48e7b022f45aba34318bcc8aad9e983bf815c25def34ad5b71cc","last_reissued_at":"2026-07-05T11:17:36.246982Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:36.246982Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TorchTitan: One-stop PyTorch native solution for production ready LLM pre-training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DC","cs.LG"],"primary_cat":"cs.CL","authors_text":"Andrew Gu, Chien-Chin Huang, Gokul Nadathur, Howard Huang, Iris Zhang, Junjie Wang, Less Wright, Sanket Purandare, Stratos Idreos, Tianyu Liu, Wanchao Liang, Wei Feng, Will Constable","submitted_at":"2024-10-09T03:26:11Z","abstract_excerpt":"The development of large language models (LLMs) has been instrumental in advancing state-of-the-art natural language processing applications. Training LLMs with billions of parameters and trillions of tokens require sophisticated distributed systems that enable composing and comparing several state-of-the-art techniques in order to efficiently scale across thousands of accelerators. However, existing solutions are complex, scattered across multiple libraries/repositories, lack interoperability, and are cumbersome to maintain. Thus, curating and empirically comparing training recipes require no"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.06511","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/2410.06511/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":"2410.06511","created_at":"2026-07-05T11:17:36.247061+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.06511v3","created_at":"2026-07-05T11:17:36.247061+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.06511","created_at":"2026-07-05T11:17:36.247061+00:00"},{"alias_kind":"pith_short_12","alias_value":"YXYZUDTEV5EO","created_at":"2026-07-05T11:17:36.247061+00:00"},{"alias_kind":"pith_short_16","alias_value":"YXYZUDTEV5EOPMBC","created_at":"2026-07-05T11:17:36.247061+00:00"},{"alias_kind":"pith_short_8","alias_value":"YXYZUDTE","created_at":"2026-07-05T11:17:36.247061+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":16,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05895","citing_title":"MatrixFSDP: communication-free matrix optimizers under ZeRO-3 parameter sharding","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2606.27153","citing_title":"DMuon: Efficient Distributed Muon Training with Near-Adam Overhead","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2607.01844","citing_title":"Mixture-of-Parallelisms: Towards Memory-Efficient Training Stack for Mixture-of-Experts Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21603","citing_title":"DynaFlow: Transparent and Flexible Intra-Device Parallelism via Programmable Operator Scheduling","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19269","citing_title":"CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19269","citing_title":"CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2512.12131","citing_title":"BOOST: BOttleneck-Optimized Scalable Training Framework for Low-Rank Large Language Models","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27263","citing_title":"Decoupling the Benefits of Subword Tokenization for Language Model Training via Byte-level Simulation","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10886","citing_title":"LoKA: Low-precision Kernel Applications for Recommendation Models At Scale","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2505.13211","citing_title":"MAGI-1: Autoregressive Video Generation at Scale","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10886","citing_title":"LoKA: Low-precision Kernel Applications for Recommendation Models At Scale","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06544","citing_title":"CCL-Bench 1.0: A Trace-Based Benchmark for LLM Infrastructure","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19241","citing_title":"UniEP: Unified Expert-Parallel MoE MegaKernel for LLM Training","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2506.15742","citing_title":"FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17550","citing_title":"Flint: Compiler Enabled Cluster-Free Design Space Exploration for Distributed ML","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27263","citing_title":"Decoupling the Benefits of Subword Tokenization for Language Model Training via Byte-level Simulation","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP","json":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP.json","graph_json":"https://pith.science/api/pith-number/YXYZUDTEV5EOPMBC6RNLUNBRRP/graph.json","events_json":"https://pith.science/api/pith-number/YXYZUDTEV5EOPMBC6RNLUNBRRP/events.json","paper":"https://pith.science/paper/YXYZUDTE"},"agent_actions":{"view_html":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP","download_json":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP.json","view_paper":"https://pith.science/paper/YXYZUDTE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.06511&json=true","fetch_graph":"https://pith.science/api/pith-number/YXYZUDTEV5EOPMBC6RNLUNBRRP/graph.json","fetch_events":"https://pith.science/api/pith-number/YXYZUDTEV5EOPMBC6RNLUNBRRP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP/action/storage_attestation","attest_author":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP/action/author_attestation","sign_citation":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP/action/citation_signature","submit_replication":"https://pith.science/pith/YXYZUDTEV5EOPMBC6RNLUNBRRP/action/replication_record"}},"created_at":"2026-07-05T11:17:36.247061+00:00","updated_at":"2026-07-05T11:17:36.247061+00:00"}