{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZB3JCTAXGPIXF6XAECPNGAPJZW","short_pith_number":"pith:ZB3JCTAX","schema_version":"1.0","canonical_sha256":"c876914c1733d172fae0209ed301e9cda65e29152b00421b39cb2fef66f76093","source":{"kind":"arxiv","id":"2507.16099","version":1},"attestation_state":"computed","paper":{"title":"TorchAO: PyTorch-Native Training-to-Serving Model Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aleksandar Samard\\v{z}i\\'c, Andrew Or, Apurva Jain, Charles David Hernandez, Christian Puhrsch, Daniel Vega-Myhre, Driss Guessous, Jesse Cai, Mark Saroufim, Supriya Rao, Thien Tran, Vasiliy Kuznetsov, Zhenrui Zheng","submitted_at":"2025-07-21T22:50:12Z","abstract_excerpt":"We present TorchAO, a PyTorch-native model optimization framework leveraging quantization and sparsity to provide an end-to-end, training-to-serving workflow for AI models. TorchAO supports a variety of popular model optimization techniques, including FP8 quantized training, quantization-aware training (QAT), post-training quantization (PTQ), and 2:4 sparsity, and leverages a novel tensor subclass abstraction to represent a variety of widely-used, backend agnostic low precision data types, including INT4, INT8, FP8, MXFP4, MXFP6, and MXFP8. TorchAO integrates closely with the broader ecosystem"},"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":"2507.16099","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-21T22:50:12Z","cross_cats_sorted":[],"title_canon_sha256":"1cb6f6f6cb3ca3387a885878cb417fb4ae73201b7274e796a8fbb151cbf4ab6c","abstract_canon_sha256":"31f3c276a9e0f6cd9e47479f4f4e262169d2ce1f40ab9e016db436a4a9672c13"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:41:02.915211Z","signature_b64":"100L9XLJn9HB8M+vYIiIlzz2KLg7I5eoDPNavnRv95HajZw3KbUBAbuzt4lF+YQlUTyTzZM+9yVVBiw26141Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c876914c1733d172fae0209ed301e9cda65e29152b00421b39cb2fef66f76093","last_reissued_at":"2026-07-05T11:41:02.914706Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:41:02.914706Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TorchAO: PyTorch-Native Training-to-Serving Model Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aleksandar Samard\\v{z}i\\'c, Andrew Or, Apurva Jain, Charles David Hernandez, Christian Puhrsch, Daniel Vega-Myhre, Driss Guessous, Jesse Cai, Mark Saroufim, Supriya Rao, Thien Tran, Vasiliy Kuznetsov, Zhenrui Zheng","submitted_at":"2025-07-21T22:50:12Z","abstract_excerpt":"We present TorchAO, a PyTorch-native model optimization framework leveraging quantization and sparsity to provide an end-to-end, training-to-serving workflow for AI models. TorchAO supports a variety of popular model optimization techniques, including FP8 quantized training, quantization-aware training (QAT), post-training quantization (PTQ), and 2:4 sparsity, and leverages a novel tensor subclass abstraction to represent a variety of widely-used, backend agnostic low precision data types, including INT4, INT8, FP8, MXFP4, MXFP6, and MXFP8. TorchAO integrates closely with the broader ecosystem"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.16099","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/2507.16099/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":"2507.16099","created_at":"2026-07-05T11:41:02.914783+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.16099v1","created_at":"2026-07-05T11:41:02.914783+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.16099","created_at":"2026-07-05T11:41:02.914783+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZB3JCTAXGPIX","created_at":"2026-07-05T11:41:02.914783+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZB3JCTAXGPIXF6XA","created_at":"2026-07-05T11:41:02.914783+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZB3JCTAX","created_at":"2026-07-05T11:41:02.914783+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21413","citing_title":"CAT-Translate: Building Compact Open-Source Models for Japanese-English Translation","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21442","citing_title":"torchtune: PyTorch native post-training library","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10886","citing_title":"LoKA: Low-precision Kernel Applications for Recommendation Models At Scale","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03420","citing_title":"Zero-Shot Quantization via Weight-Space Arithmetic","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10886","citing_title":"LoKA: Low-precision Kernel Applications for Recommendation Models At Scale","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15416","citing_title":"StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW","json":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW.json","graph_json":"https://pith.science/api/pith-number/ZB3JCTAXGPIXF6XAECPNGAPJZW/graph.json","events_json":"https://pith.science/api/pith-number/ZB3JCTAXGPIXF6XAECPNGAPJZW/events.json","paper":"https://pith.science/paper/ZB3JCTAX"},"agent_actions":{"view_html":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW","download_json":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW.json","view_paper":"https://pith.science/paper/ZB3JCTAX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.16099&json=true","fetch_graph":"https://pith.science/api/pith-number/ZB3JCTAXGPIXF6XAECPNGAPJZW/graph.json","fetch_events":"https://pith.science/api/pith-number/ZB3JCTAXGPIXF6XAECPNGAPJZW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW/action/storage_attestation","attest_author":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW/action/author_attestation","sign_citation":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW/action/citation_signature","submit_replication":"https://pith.science/pith/ZB3JCTAXGPIXF6XAECPNGAPJZW/action/replication_record"}},"created_at":"2026-07-05T11:41:02.914783+00:00","updated_at":"2026-07-05T11:41:02.914783+00:00"}