{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:JPKXLFDROQIYCBQG6LWOMOOJAE","short_pith_number":"pith:JPKXLFDR","canonical_record":{"source":{"id":"2310.10837","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-16T21:23:16Z","cross_cats_sorted":["cs.NE"],"title_canon_sha256":"dd28295ef045e16e512b74e472d8fbd4bb9329e383c3c544bf793e052343985b","abstract_canon_sha256":"23a6ba15120ebe570174ec2dc648fed9fe873ce1c04a1fcf7ab34d57f85254c1"},"schema_version":"1.0"},"canonical_sha256":"4bd57594717411810606f2ece639c901327d1bb998241fb2a4d79fec74810dd1","source":{"kind":"arxiv","id":"2310.10837","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.10837","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"arxiv_version","alias_value":"2310.10837v3","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.10837","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"pith_short_12","alias_value":"JPKXLFDROQIY","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"pith_short_16","alias_value":"JPKXLFDROQIYCBQG","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"pith_short_8","alias_value":"JPKXLFDR","created_at":"2026-07-05T07:14:57Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:JPKXLFDROQIYCBQG6LWOMOOJAE","target":"record","payload":{"canonical_record":{"source":{"id":"2310.10837","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-16T21:23:16Z","cross_cats_sorted":["cs.NE"],"title_canon_sha256":"dd28295ef045e16e512b74e472d8fbd4bb9329e383c3c544bf793e052343985b","abstract_canon_sha256":"23a6ba15120ebe570174ec2dc648fed9fe873ce1c04a1fcf7ab34d57f85254c1"},"schema_version":"1.0"},"canonical_sha256":"4bd57594717411810606f2ece639c901327d1bb998241fb2a4d79fec74810dd1","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:14:57.388869Z","signature_b64":"U1gT3Ygm+ufWRCtoZt8UeTrYZq3XRgEHedBz1sJRVlQyboO4ochoSmUjSVfKJQKUQMoexomE5RMMhphBjjj3CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4bd57594717411810606f2ece639c901327d1bb998241fb2a4d79fec74810dd1","last_reissued_at":"2026-07-05T07:14:57.388450Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:14:57.388450Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2310.10837","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:14:57Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lH+OklkkU2ysCIJsQMtjzrkukastSmBptft/Mmr8V7XCzjmHxQ3q8z9QN/Lz+zKr8EcMcadrFcHvz6b/ARvrCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T01:13:48.707717Z"},"content_sha256":"f447a6407002b5936ff319cfce14b77400895eb6fea28f72ab96e4d0391d82f8","schema_version":"1.0","event_id":"sha256:f447a6407002b5936ff319cfce14b77400895eb6fea28f72ab96e4d0391d82f8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:JPKXLFDROQIYCBQG6LWOMOOJAE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Approximating Two-Layer Feedforward Networks for Efficient Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.NE"],"primary_cat":"cs.LG","authors_text":"J\\\"urgen Schmidhuber, Kazuki Irie, R\\'obert Csord\\'as","submitted_at":"2023-10-16T21:23:16Z","abstract_excerpt":"How to reduce compute and memory requirements of neural networks (NNs) without sacrificing performance? Many recent works use sparse Mixtures of Experts (MoEs) to build resource-efficient large language models (LMs). Here we introduce several novel perspectives on MoEs, presenting a general framework that unifies various methods to approximate two-layer NNs (e.g., feedforward blocks of Transformers), including product-key memories (PKMs). Leveraging insights from this framework, we propose methods to improve both MoEs and PKMs. Unlike prior work that compares MoEs with dense baselines under th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.10837","kind":"arxiv","version":3},"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/2310.10837/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":"2026-07-05T07:14:57Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"c2iPn7oBgwqVDWNlgdXRIL0vOYxW04cWTnY/m54jF5+3/CSaf7BRRcmcQo5hPGCey62RF/KNVCZtZU32y47tCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T01:13:48.708336Z"},"content_sha256":"6f5f8517307c1631d15c7e7c4d0c113d001d0d8c40f260435c3eb4993b4c473c","schema_version":"1.0","event_id":"sha256:6f5f8517307c1631d15c7e7c4d0c113d001d0d8c40f260435c3eb4993b4c473c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/JPKXLFDROQIYCBQG6LWOMOOJAE/bundle.json","state_url":"https://pith.science/pith/JPKXLFDROQIYCBQG6LWOMOOJAE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/JPKXLFDROQIYCBQG6LWOMOOJAE/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":"2026-08-06T01:13:48Z","links":{"resolver":"https://pith.science/pith/JPKXLFDROQIYCBQG6LWOMOOJAE","bundle":"https://pith.science/pith/JPKXLFDROQIYCBQG6LWOMOOJAE/bundle.json","state":"https://pith.science/pith/JPKXLFDROQIYCBQG6LWOMOOJAE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/JPKXLFDROQIYCBQG6LWOMOOJAE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:JPKXLFDROQIYCBQG6LWOMOOJAE","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":"23a6ba15120ebe570174ec2dc648fed9fe873ce1c04a1fcf7ab34d57f85254c1","cross_cats_sorted":["cs.NE"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-16T21:23:16Z","title_canon_sha256":"dd28295ef045e16e512b74e472d8fbd4bb9329e383c3c544bf793e052343985b"},"schema_version":"1.0","source":{"id":"2310.10837","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.10837","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"arxiv_version","alias_value":"2310.10837v3","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.10837","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"pith_short_12","alias_value":"JPKXLFDROQIY","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"pith_short_16","alias_value":"JPKXLFDROQIYCBQG","created_at":"2026-07-05T07:14:57Z"},{"alias_kind":"pith_short_8","alias_value":"JPKXLFDR","created_at":"2026-07-05T07:14:57Z"}],"graph_snapshots":[{"event_id":"sha256:6f5f8517307c1631d15c7e7c4d0c113d001d0d8c40f260435c3eb4993b4c473c","target":"graph","created_at":"2026-07-05T07:14:57Z","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/2310.10837/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"How to reduce compute and memory requirements of neural networks (NNs) without sacrificing performance? Many recent works use sparse Mixtures of Experts (MoEs) to build resource-efficient large language models (LMs). Here we introduce several novel perspectives on MoEs, presenting a general framework that unifies various methods to approximate two-layer NNs (e.g., feedforward blocks of Transformers), including product-key memories (PKMs). Leveraging insights from this framework, we propose methods to improve both MoEs and PKMs. Unlike prior work that compares MoEs with dense baselines under th","authors_text":"J\\\"urgen Schmidhuber, Kazuki Irie, R\\'obert Csord\\'as","cross_cats":["cs.NE"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-16T21:23:16Z","title":"Approximating Two-Layer Feedforward Networks for Efficient Transformers"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.10837","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:f447a6407002b5936ff319cfce14b77400895eb6fea28f72ab96e4d0391d82f8","target":"record","created_at":"2026-07-05T07:14:57Z","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":"23a6ba15120ebe570174ec2dc648fed9fe873ce1c04a1fcf7ab34d57f85254c1","cross_cats_sorted":["cs.NE"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-16T21:23:16Z","title_canon_sha256":"dd28295ef045e16e512b74e472d8fbd4bb9329e383c3c544bf793e052343985b"},"schema_version":"1.0","source":{"id":"2310.10837","kind":"arxiv","version":3}},"canonical_sha256":"4bd57594717411810606f2ece639c901327d1bb998241fb2a4d79fec74810dd1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4bd57594717411810606f2ece639c901327d1bb998241fb2a4d79fec74810dd1","first_computed_at":"2026-07-05T07:14:57.388450Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:14:57.388450Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"U1gT3Ygm+ufWRCtoZt8UeTrYZq3XRgEHedBz1sJRVlQyboO4ochoSmUjSVfKJQKUQMoexomE5RMMhphBjjj3CQ==","signature_status":"signed_v1","signed_at":"2026-07-05T07:14:57.388869Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.10837","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f447a6407002b5936ff319cfce14b77400895eb6fea28f72ab96e4d0391d82f8","sha256:6f5f8517307c1631d15c7e7c4d0c113d001d0d8c40f260435c3eb4993b4c473c"],"state_sha256":"71dabd62ee298ef69c645013cdef8f3c2c5c29acf51aedc515a44c3d574e3d96"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oJjxXTKzok31n70ufY1jaNPtyVCfFGCDfcJ5MdztqC7E2hhh3Te0+xKodqyuZWAzfusImsC6YsEYO+qIyLvCCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T01:13:48.713996Z","bundle_sha256":"36af3cad23d11c2e0026943a70d6cf40b6d241bcd425520fcf6180ba43aff69e"}}