{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:5GLYRQWISBMTTF3GN5S5FKBQ66","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":"fb145049edde77775fb72af8b6b40a4ae2381029475cb35bc5fca05534044295","cross_cats_sorted":["cs.AI","math.OC"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-20T00:10:35Z","title_canon_sha256":"e23700cd447f3fb289457403c1de9e4e4ccd2b6cd75f0238eda93186941fe2ea"},"schema_version":"1.0","source":{"id":"2506.16659","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.16659","created_at":"2026-05-22T01:03:43Z"},{"alias_kind":"arxiv_version","alias_value":"2506.16659v3","created_at":"2026-05-22T01:03:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.16659","created_at":"2026-05-22T01:03:43Z"},{"alias_kind":"pith_short_12","alias_value":"5GLYRQWISBMT","created_at":"2026-05-22T01:03:43Z"},{"alias_kind":"pith_short_16","alias_value":"5GLYRQWISBMTTF3G","created_at":"2026-05-22T01:03:43Z"},{"alias_kind":"pith_short_8","alias_value":"5GLYRQWI","created_at":"2026-05-22T01:03:43Z"}],"graph_snapshots":[{"event_id":"sha256:c01f01d840fa980beb96215a775c4e234c768a15fa99fd81d7ba249a0e2217eb","target":"graph","created_at":"2026-05-22T01:03:43Z","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/2506.16659/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Training large language models (LLMs) relies on adaptive optimizers such as Adam, which introduce extra operations and require significantly more memory to maintain first- and second-order moments than SGD. While recent works such as GaLore, Fira and APOLLO have proposed state-compressed memory-efficient variants, a fundamental question remains: What are the minimum modifications to plain SGD needed to match state-of-the-art pretraining performance? We systematically investigate this question using a bottom-up approach, and identify two simple yet highly (memory- and compute-) efficient techni","authors_text":"Andi Han, Athanasios Glentis, Jiaxiang Li, Mingyi Hong","cross_cats":["cs.AI","math.OC"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-20T00:10:35Z","title":"Memory-Efficient LLM Pretraining via Minimalist Optimizer Design"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.16659","kind":"arxiv","version":3},"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":"sha256:677888e35a1adda19e7342a556c189715a23c09a6641b300d4275042fb84291c","target":"record","created_at":"2026-05-22T01:03:43Z","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":"fb145049edde77775fb72af8b6b40a4ae2381029475cb35bc5fca05534044295","cross_cats_sorted":["cs.AI","math.OC"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-20T00:10:35Z","title_canon_sha256":"e23700cd447f3fb289457403c1de9e4e4ccd2b6cd75f0238eda93186941fe2ea"},"schema_version":"1.0","source":{"id":"2506.16659","kind":"arxiv","version":3}},"canonical_sha256":"e99788c2c890593997666f65d2a830f79dc294613e3e1f0b9a257e5a4c8ad390","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e99788c2c890593997666f65d2a830f79dc294613e3e1f0b9a257e5a4c8ad390","first_computed_at":"2026-05-22T01:03:43.626630Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-22T01:03:43.626630Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"TnpRBpAGahUvU9uiv00WD6hKQmwHKvxcH4WolPu/mcTPfCN0IkEqBYw2MY2+OKcHk1mB7EEZHOgIJHhW0ianAA==","signature_status":"signed_v1","signed_at":"2026-05-22T01:03:43.627522Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.16659","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:677888e35a1adda19e7342a556c189715a23c09a6641b300d4275042fb84291c","sha256:c01f01d840fa980beb96215a775c4e234c768a15fa99fd81d7ba249a0e2217eb"],"state_sha256":"82ae710ee577f0d1d34204e4f458b3202f07ef9ebf7815fd884ce462773b8c7a"}