{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:55FVKBP5FC5QCY5ESL6ZJD5EKQ","short_pith_number":"pith:55FVKBP5","canonical_record":{"source":{"id":"2310.18313","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T17:59:51Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3c811e3704ba7d1eeb1ea9b3f29d7dd01add521a6285c6b2b93f5c0e31aa8fe7","abstract_canon_sha256":"4cd97b51d6c32b9747c4d556f88991c8ec9d0aa28a1c987a5bfbad88b5eb685c"},"schema_version":"1.0"},"canonical_sha256":"ef4b5505fd28bb0163a492fd948fa45427a37800739d4c845a730ed18d4b04b4","source":{"kind":"arxiv","id":"2310.18313","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.18313","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"arxiv_version","alias_value":"2310.18313v2","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.18313","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"pith_short_12","alias_value":"55FVKBP5FC5Q","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"pith_short_16","alias_value":"55FVKBP5FC5QCY5E","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"pith_short_8","alias_value":"55FVKBP5","created_at":"2026-07-05T07:25:47Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:55FVKBP5FC5QCY5ESL6ZJD5EKQ","target":"record","payload":{"canonical_record":{"source":{"id":"2310.18313","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T17:59:51Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3c811e3704ba7d1eeb1ea9b3f29d7dd01add521a6285c6b2b93f5c0e31aa8fe7","abstract_canon_sha256":"4cd97b51d6c32b9747c4d556f88991c8ec9d0aa28a1c987a5bfbad88b5eb685c"},"schema_version":"1.0"},"canonical_sha256":"ef4b5505fd28bb0163a492fd948fa45427a37800739d4c845a730ed18d4b04b4","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:25:47.453817Z","signature_b64":"srsiYgSZR+jTqw0vXxdlkEGDs1YAxmDShWEX636vWiiOvA7qdolEP1k2hDxzJxk2FKdorZ8fhUyhXm9WziwsCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ef4b5505fd28bb0163a492fd948fa45427a37800739d4c845a730ed18d4b04b4","last_reissued_at":"2026-07-05T07:25:47.453367Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:25:47.453367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2310.18313","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:25:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7cOCWD7PRN5R1wb0La/0HudoI08Ww2kfmcw+kOtXxZVtHa2zVE/tAKlgMhA58ZYn8QiccJ8y5jje1BBYdzvoCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T10:21:38.015089Z"},"content_sha256":"778d4aed8cb98d69a0f577ff1bb97fe3598001992fb0b54dd839d26d10998aab","schema_version":"1.0","event_id":"sha256:778d4aed8cb98d69a0f577ff1bb97fe3598001992fb0b54dd839d26d10998aab"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:55FVKBP5FC5QCY5ESL6ZJD5EKQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"FP8-LM: Training FP8 Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Bolin Ni, Chen Li, Guoshuai Zhao, Han Hu, Houwen Peng, Jia Ning, Jingcheng Hu, Joe Chau, Kan Wu, Miaosen Zhang, Peng Cheng, Ruihang Li, Ruizhe Wang, Shuguang Liu, Yifan Xiong, Yixuan Wei, Yuxiang Yang, Ze Liu, Zheng Zhang, Ziyue Yang","submitted_at":"2023-10-27T17:59:51Z","abstract_excerpt":"In this paper, we explore FP8 low-bit data formats for efficient training of large language models (LLMs). Our key insight is that most variables, such as gradients and optimizer states, in LLM training can employ low-precision data formats without compromising model accuracy and requiring no changes to hyper-parameters. Specifically, we propose a new FP8 automatic mixed-precision framework for training LLMs. This framework offers three levels of FP8 utilization to streamline mixed-precision and distributed parallel training for LLMs. It gradually incorporates 8-bit gradients, optimizer states"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.18313","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/2310.18313/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:25:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FN32IYYiv8XQhxV/SEh5VT7Wuvkl+lvfPaElUFhQHBtpqfPDuHEtEp03Vl8blujGvg74yegybeSIFGQJtHLiCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T10:21:38.016105Z"},"content_sha256":"690421bd3d35b6acb14fbb5b8a08f834dee2f23c4b4b047c3a92ab022bfb5faa","schema_version":"1.0","event_id":"sha256:690421bd3d35b6acb14fbb5b8a08f834dee2f23c4b4b047c3a92ab022bfb5faa"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ/bundle.json","state_url":"https://pith.science/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-04T10:21:38Z","links":{"resolver":"https://pith.science/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ","bundle":"https://pith.science/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ/bundle.json","state":"https://pith.science/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/55FVKBP5FC5QCY5ESL6ZJD5EKQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:55FVKBP5FC5QCY5ESL6ZJD5EKQ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"4cd97b51d6c32b9747c4d556f88991c8ec9d0aa28a1c987a5bfbad88b5eb685c","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T17:59:51Z","title_canon_sha256":"3c811e3704ba7d1eeb1ea9b3f29d7dd01add521a6285c6b2b93f5c0e31aa8fe7"},"schema_version":"1.0","source":{"id":"2310.18313","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.18313","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"arxiv_version","alias_value":"2310.18313v2","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.18313","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"pith_short_12","alias_value":"55FVKBP5FC5Q","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"pith_short_16","alias_value":"55FVKBP5FC5QCY5E","created_at":"2026-07-05T07:25:47Z"},{"alias_kind":"pith_short_8","alias_value":"55FVKBP5","created_at":"2026-07-05T07:25:47Z"}],"graph_snapshots":[{"event_id":"sha256:690421bd3d35b6acb14fbb5b8a08f834dee2f23c4b4b047c3a92ab022bfb5faa","target":"graph","created_at":"2026-07-05T07:25:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2310.18313/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In this paper, we explore FP8 low-bit data formats for efficient training of large language models (LLMs). Our key insight is that most variables, such as gradients and optimizer states, in LLM training can employ low-precision data formats without compromising model accuracy and requiring no changes to hyper-parameters. Specifically, we propose a new FP8 automatic mixed-precision framework for training LLMs. This framework offers three levels of FP8 utilization to streamline mixed-precision and distributed parallel training for LLMs. It gradually incorporates 8-bit gradients, optimizer states","authors_text":"Bolin Ni, Chen Li, Guoshuai Zhao, Han Hu, Houwen Peng, Jia Ning, Jingcheng Hu, Joe Chau, Kan Wu, Miaosen Zhang, Peng Cheng, Ruihang Li, Ruizhe Wang, Shuguang Liu, Yifan Xiong, Yixuan Wei, Yuxiang Yang, Ze Liu, Zheng Zhang, Ziyue Yang","cross_cats":["cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T17:59:51Z","title":"FP8-LM: Training FP8 Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.18313","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:778d4aed8cb98d69a0f577ff1bb97fe3598001992fb0b54dd839d26d10998aab","target":"record","created_at":"2026-07-05T07:25:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"4cd97b51d6c32b9747c4d556f88991c8ec9d0aa28a1c987a5bfbad88b5eb685c","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T17:59:51Z","title_canon_sha256":"3c811e3704ba7d1eeb1ea9b3f29d7dd01add521a6285c6b2b93f5c0e31aa8fe7"},"schema_version":"1.0","source":{"id":"2310.18313","kind":"arxiv","version":2}},"canonical_sha256":"ef4b5505fd28bb0163a492fd948fa45427a37800739d4c845a730ed18d4b04b4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ef4b5505fd28bb0163a492fd948fa45427a37800739d4c845a730ed18d4b04b4","first_computed_at":"2026-07-05T07:25:47.453367Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:25:47.453367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"srsiYgSZR+jTqw0vXxdlkEGDs1YAxmDShWEX636vWiiOvA7qdolEP1k2hDxzJxk2FKdorZ8fhUyhXm9WziwsCw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:25:47.453817Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.18313","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:778d4aed8cb98d69a0f577ff1bb97fe3598001992fb0b54dd839d26d10998aab","sha256:690421bd3d35b6acb14fbb5b8a08f834dee2f23c4b4b047c3a92ab022bfb5faa"],"state_sha256":"1bc6395f1d9f0b390daa1060034e8fa9b2ecac1d305fc06d3eed3207ec1a94f0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0ASm02zv1xz8yB6bO0F4gp7jP7VpHrfXDp2Wr4gUx3ewMMJAnOWwPgyjMjyoxJwGwc8jvwzeA9KThUmHpcR9Bg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T10:21:38.022096Z","bundle_sha256":"ae40837ba5f0578ed3d52c8f1c6cc5fdb97a9dc4180404f3ce9922ca5437ab7a"}}