{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:SGJNRTRPIQ367KL2GZTNRCN2ZB","short_pith_number":"pith:SGJNRTRP","canonical_record":{"source":{"id":"2508.20577","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-28T09:14:23Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"e0fd49c2a5ff8b7b5c0f803344fd5a6a13d64a14a1276abbb421ef66e0f9d3bd","abstract_canon_sha256":"79d57b329ece3afae741e2d6c08b7696a31471f940c7c71c05d98401e02f8eed"},"schema_version":"1.0"},"canonical_sha256":"9192d8ce2f4437efa97a3666d889bac87ecca379b16820cf168f85d4ec51e0d9","source":{"kind":"arxiv","id":"2508.20577","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.20577","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"arxiv_version","alias_value":"2508.20577v1","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.20577","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"pith_short_12","alias_value":"SGJNRTRPIQ36","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"pith_short_16","alias_value":"SGJNRTRPIQ367KL2","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"pith_short_8","alias_value":"SGJNRTRP","created_at":"2026-07-05T12:00:55Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:SGJNRTRPIQ367KL2GZTNRCN2ZB","target":"record","payload":{"canonical_record":{"source":{"id":"2508.20577","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-28T09:14:23Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"e0fd49c2a5ff8b7b5c0f803344fd5a6a13d64a14a1276abbb421ef66e0f9d3bd","abstract_canon_sha256":"79d57b329ece3afae741e2d6c08b7696a31471f940c7c71c05d98401e02f8eed"},"schema_version":"1.0"},"canonical_sha256":"9192d8ce2f4437efa97a3666d889bac87ecca379b16820cf168f85d4ec51e0d9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:55.236210Z","signature_b64":"uWFy84SSDIehkFVONi95VUfrN/ohsNt0eWMtExImzryDhyOq/YIN3wgbP73/Y2KLG7RzJLS3rgemlhi9r7F1Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9192d8ce2f4437efa97a3666d889bac87ecca379b16820cf168f85d4ec51e0d9","last_reissued_at":"2026-07-05T12:00:55.235659Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:55.235659Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2508.20577","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T12:00:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"al6ekFCNhbJ8JyvWaojWBSBgzDpnzp/JBVcowGzyOT93JaAPUfQXMmgZntHSZGna74gIuKsJ3F4GOv9NNVNpBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T20:45:48.668594Z"},"content_sha256":"f0f12680c495f88a1d99d3b6a6f0f60401fa9a1ea7bb800361aa4baa6db082b8","schema_version":"1.0","event_id":"sha256:f0f12680c495f88a1d99d3b6a6f0f60401fa9a1ea7bb800361aa4baa6db082b8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:SGJNRTRPIQ367KL2GZTNRCN2ZB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Yang Luo, Yang You, Yong Liu, Zangwei Zheng, Ziheng Qin, Zirui Zhu","submitted_at":"2025-08-28T09:14:23Z","abstract_excerpt":"Large-batch training has become a cornerstone in accelerating the training of deep neural networks, yet it poses challenges in optimization and generalization. Existing optimizers like AdamW present performance degradation during language models' large-batch training, due to the information bottleneck in attention layers caused by the sharp increase of max attention logit. While the LAMB optimizer partially addresses this issue, some attention layers still face this issue. The reason is that $l_2$-norm-based trust ratios in LAMB are less effective in directly influencing the max value of query"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.20577","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/2508.20577/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":"2026-07-05T12:00:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zo9S9qGPsSZ/YLt7Ka5njTRL4WHOMxMys5Zwddosvv4VApEto6806IRProwykNvJW1bX5ymxS+fGtYARPq8VCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T20:45:48.669544Z"},"content_sha256":"0c719ccc5f704fa0c8169d722ef86befe93a706eab748b776d83ea2328dce04e","schema_version":"1.0","event_id":"sha256:0c719ccc5f704fa0c8169d722ef86befe93a706eab748b776d83ea2328dce04e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB/bundle.json","state_url":"https://pith.science/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB/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":"2026-08-17T20:45:48Z","links":{"resolver":"https://pith.science/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB","bundle":"https://pith.science/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB/bundle.json","state":"https://pith.science/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SGJNRTRPIQ367KL2GZTNRCN2ZB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:SGJNRTRPIQ367KL2GZTNRCN2ZB","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":"79d57b329ece3afae741e2d6c08b7696a31471f940c7c71c05d98401e02f8eed","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-28T09:14:23Z","title_canon_sha256":"e0fd49c2a5ff8b7b5c0f803344fd5a6a13d64a14a1276abbb421ef66e0f9d3bd"},"schema_version":"1.0","source":{"id":"2508.20577","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.20577","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"arxiv_version","alias_value":"2508.20577v1","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.20577","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"pith_short_12","alias_value":"SGJNRTRPIQ36","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"pith_short_16","alias_value":"SGJNRTRPIQ367KL2","created_at":"2026-07-05T12:00:55Z"},{"alias_kind":"pith_short_8","alias_value":"SGJNRTRP","created_at":"2026-07-05T12:00:55Z"}],"graph_snapshots":[{"event_id":"sha256:0c719ccc5f704fa0c8169d722ef86befe93a706eab748b776d83ea2328dce04e","target":"graph","created_at":"2026-07-05T12:00:55Z","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/2508.20577/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large-batch training has become a cornerstone in accelerating the training of deep neural networks, yet it poses challenges in optimization and generalization. Existing optimizers like AdamW present performance degradation during language models' large-batch training, due to the information bottleneck in attention layers caused by the sharp increase of max attention logit. While the LAMB optimizer partially addresses this issue, some attention layers still face this issue. The reason is that $l_2$-norm-based trust ratios in LAMB are less effective in directly influencing the max value of query","authors_text":"Yang Luo, Yang You, Yong Liu, Zangwei Zheng, Ziheng Qin, Zirui Zhu","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-28T09:14:23Z","title":"MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.20577","kind":"arxiv","version":1},"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:f0f12680c495f88a1d99d3b6a6f0f60401fa9a1ea7bb800361aa4baa6db082b8","target":"record","created_at":"2026-07-05T12:00:55Z","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":"79d57b329ece3afae741e2d6c08b7696a31471f940c7c71c05d98401e02f8eed","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-28T09:14:23Z","title_canon_sha256":"e0fd49c2a5ff8b7b5c0f803344fd5a6a13d64a14a1276abbb421ef66e0f9d3bd"},"schema_version":"1.0","source":{"id":"2508.20577","kind":"arxiv","version":1}},"canonical_sha256":"9192d8ce2f4437efa97a3666d889bac87ecca379b16820cf168f85d4ec51e0d9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9192d8ce2f4437efa97a3666d889bac87ecca379b16820cf168f85d4ec51e0d9","first_computed_at":"2026-07-05T12:00:55.235659Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T12:00:55.235659Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"uWFy84SSDIehkFVONi95VUfrN/ohsNt0eWMtExImzryDhyOq/YIN3wgbP73/Y2KLG7RzJLS3rgemlhi9r7F1Aw==","signature_status":"signed_v1","signed_at":"2026-07-05T12:00:55.236210Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.20577","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f0f12680c495f88a1d99d3b6a6f0f60401fa9a1ea7bb800361aa4baa6db082b8","sha256:0c719ccc5f704fa0c8169d722ef86befe93a706eab748b776d83ea2328dce04e"],"state_sha256":"c1d6f6807f55bde11a37e8a81b2cf4a32e239a45e0d60315ee838fc0be1d076f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xCPu8TUIvAODO8tFfLEJN9sojhRVBHY8fkUHuiwMOaKV5BLfC8ZoGVP4VSjPVP1hRn6EKmQsMMHOq11aAcyyBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T20:45:48.870800Z","bundle_sha256":"0f1d5b84c057233d70d18b22e590f2596aa291d166f569aee783209156ddcd1f"}}