{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:CPP6TZDTJN4B5GMLSQF6XRRZPZ","short_pith_number":"pith:CPP6TZDT","canonical_record":{"source":{"id":"2409.03137","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-09-05T00:13:16Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"dff58c3f9d487cd35a12838cafb1931529735162b4fe6c37757e73036991a6be","abstract_canon_sha256":"3018f81732e1898f4a2a0ad2de8bba2113f5997e22a94d69560007d92c71e7b3"},"schema_version":"1.0"},"canonical_sha256":"13dfe9e4734b781e998b940bebc6397e73607139045ccb40f566c252599f8787","source":{"kind":"arxiv","id":"2409.03137","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2409.03137","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"arxiv_version","alias_value":"2409.03137v2","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.03137","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"pith_short_12","alias_value":"CPP6TZDTJN4B","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"pith_short_16","alias_value":"CPP6TZDTJN4B5GML","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"pith_short_8","alias_value":"CPP6TZDT","created_at":"2026-07-05T09:13:07Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:CPP6TZDTJN4B5GMLSQF6XRRZPZ","target":"record","payload":{"canonical_record":{"source":{"id":"2409.03137","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-09-05T00:13:16Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"dff58c3f9d487cd35a12838cafb1931529735162b4fe6c37757e73036991a6be","abstract_canon_sha256":"3018f81732e1898f4a2a0ad2de8bba2113f5997e22a94d69560007d92c71e7b3"},"schema_version":"1.0"},"canonical_sha256":"13dfe9e4734b781e998b940bebc6397e73607139045ccb40f566c252599f8787","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:13:07.102890Z","signature_b64":"ZgFSxGhsZnoIbXdn1y4Y5WlaCe3VU7TDSXo/c/4m8CeF3QcsuZJ24Wswhg/Q4fReQFTvXih5fe4Vd71QRPTnAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"13dfe9e4734b781e998b940bebc6397e73607139045ccb40f566c252599f8787","last_reissued_at":"2026-07-05T09:13:07.102271Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:13:07.102271Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2409.03137","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":"2026-07-05T09:13:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uskTQnCkyICn/myj4v+RNDtxbbwm9F6fkwk1lfFeSvDno4BJnPSr57IQ029Ti75LCT6JfyfhyMi0q+6e5t4UBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T13:16:48.062039Z"},"content_sha256":"4ff98ff6add1c5db15c1c29a4cd83941c08a0164e3adb0cdae7aa238d4f6b087","schema_version":"1.0","event_id":"sha256:4ff98ff6add1c5db15c1c29a4cd83941c08a0164e3adb0cdae7aa238d4f6b087"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:CPP6TZDTJN4B5GMLSQF6XRRZPZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"The AdEMAMix Optimizer: Better, Faster, Older","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"David Grangier, Matteo Pagliardini, Pierre Ablin","submitted_at":"2024-09-05T00:13:16Z","abstract_excerpt":"Momentum based optimizers are central to a wide range of machine learning applications. These typically rely on an Exponential Moving Average (EMA) of gradients, which decays exponentially the present contribution of older gradients. This accounts for gradients being local linear approximations which lose their relevance as the iterate moves along the loss landscape. This work questions the use of a single EMA to accumulate past gradients and empirically demonstrates how this choice can be sub-optimal: a single EMA cannot simultaneously give a high weight to the immediate past, and a non-negli"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.03137","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/2409.03137/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":"2026-07-05T09:13:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uldzPD8VzwjHV9fkLpjGsOA8dwS4k5SiYs+kNRKhpjbSc62kBCXUv7Voei2jra/45z+YNrVYCdRSOycEkOwBBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T13:16:48.062562Z"},"content_sha256":"1198472910783d0190ae7f55785cbaaabae1ca4cd5bbad97c670ffc36528fd93","schema_version":"1.0","event_id":"sha256:1198472910783d0190ae7f55785cbaaabae1ca4cd5bbad97c670ffc36528fd93"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ/bundle.json","state_url":"https://pith.science/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ/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":"2026-08-05T13:16:48Z","links":{"resolver":"https://pith.science/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ","bundle":"https://pith.science/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ/bundle.json","state":"https://pith.science/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/CPP6TZDTJN4B5GMLSQF6XRRZPZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:CPP6TZDTJN4B5GMLSQF6XRRZPZ","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":"3018f81732e1898f4a2a0ad2de8bba2113f5997e22a94d69560007d92c71e7b3","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-09-05T00:13:16Z","title_canon_sha256":"dff58c3f9d487cd35a12838cafb1931529735162b4fe6c37757e73036991a6be"},"schema_version":"1.0","source":{"id":"2409.03137","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2409.03137","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"arxiv_version","alias_value":"2409.03137v2","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.03137","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"pith_short_12","alias_value":"CPP6TZDTJN4B","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"pith_short_16","alias_value":"CPP6TZDTJN4B5GML","created_at":"2026-07-05T09:13:07Z"},{"alias_kind":"pith_short_8","alias_value":"CPP6TZDT","created_at":"2026-07-05T09:13:07Z"}],"graph_snapshots":[{"event_id":"sha256:1198472910783d0190ae7f55785cbaaabae1ca4cd5bbad97c670ffc36528fd93","target":"graph","created_at":"2026-07-05T09:13:07Z","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/2409.03137/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Momentum based optimizers are central to a wide range of machine learning applications. These typically rely on an Exponential Moving Average (EMA) of gradients, which decays exponentially the present contribution of older gradients. This accounts for gradients being local linear approximations which lose their relevance as the iterate moves along the loss landscape. This work questions the use of a single EMA to accumulate past gradients and empirically demonstrates how this choice can be sub-optimal: a single EMA cannot simultaneously give a high weight to the immediate past, and a non-negli","authors_text":"David Grangier, Matteo Pagliardini, Pierre Ablin","cross_cats":["stat.ML"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-09-05T00:13:16Z","title":"The AdEMAMix Optimizer: Better, Faster, Older"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.03137","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":"sha256:4ff98ff6add1c5db15c1c29a4cd83941c08a0164e3adb0cdae7aa238d4f6b087","target":"record","created_at":"2026-07-05T09:13:07Z","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":"3018f81732e1898f4a2a0ad2de8bba2113f5997e22a94d69560007d92c71e7b3","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-09-05T00:13:16Z","title_canon_sha256":"dff58c3f9d487cd35a12838cafb1931529735162b4fe6c37757e73036991a6be"},"schema_version":"1.0","source":{"id":"2409.03137","kind":"arxiv","version":2}},"canonical_sha256":"13dfe9e4734b781e998b940bebc6397e73607139045ccb40f566c252599f8787","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"13dfe9e4734b781e998b940bebc6397e73607139045ccb40f566c252599f8787","first_computed_at":"2026-07-05T09:13:07.102271Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:13:07.102271Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ZgFSxGhsZnoIbXdn1y4Y5WlaCe3VU7TDSXo/c/4m8CeF3QcsuZJ24Wswhg/Q4fReQFTvXih5fe4Vd71QRPTnAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:13:07.102890Z","signed_message":"canonical_sha256_bytes"},"source_id":"2409.03137","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4ff98ff6add1c5db15c1c29a4cd83941c08a0164e3adb0cdae7aa238d4f6b087","sha256:1198472910783d0190ae7f55785cbaaabae1ca4cd5bbad97c670ffc36528fd93"],"state_sha256":"f48d36388f16efdf54c370282c96d0b31691ec8bf32d97d9549faf3beaa1ae8e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"E3uLJgdN3G6HcwSD2cg8vLN1ko1VCIdeh4SQ0LiGcD1EHJ7VeLIjN9DxKpLZsONGeUtxb35JX6mlpeEQUG6ACQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T13:16:48.065922Z","bundle_sha256":"ce4d62fe7ff8a4999f306093b58a7e5ffa362e6fbe5c7c806ec8c15e816f5d9e"}}