{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:3L3YZLUTYJCM6TMSOOTHHGYVAQ","short_pith_number":"pith:3L3YZLUT","canonical_record":{"source":{"id":"2504.05586","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-08T00:49:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cce9f0ef37c70e6efbafee2cbd4fd50c5e4e3485b5e25517f4f50f0424154b16","abstract_canon_sha256":"acfde7fe45a829cfbf3ee48a7bbde2cd258f462191cbf28439ea1af34ad88b35"},"schema_version":"1.0"},"canonical_sha256":"daf78cae93c244cf4d9273a6739b15042b65b6604ba76792faed59fe4df19285","source":{"kind":"arxiv","id":"2504.05586","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.05586","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"arxiv_version","alias_value":"2504.05586v2","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.05586","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"pith_short_12","alias_value":"3L3YZLUTYJCM","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"pith_short_16","alias_value":"3L3YZLUTYJCM6TMS","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"pith_short_8","alias_value":"3L3YZLUT","created_at":"2026-07-05T10:47:02Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:3L3YZLUTYJCM6TMSOOTHHGYVAQ","target":"record","payload":{"canonical_record":{"source":{"id":"2504.05586","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-08T00:49:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cce9f0ef37c70e6efbafee2cbd4fd50c5e4e3485b5e25517f4f50f0424154b16","abstract_canon_sha256":"acfde7fe45a829cfbf3ee48a7bbde2cd258f462191cbf28439ea1af34ad88b35"},"schema_version":"1.0"},"canonical_sha256":"daf78cae93c244cf4d9273a6739b15042b65b6604ba76792faed59fe4df19285","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:47:02.591190Z","signature_b64":"kowCY/RP84fyjPB7rlUuVBjG3AImKxLpGMIh3FrakqAZFgA2g1B1eNwTCGQFcirWxPaP7bZhQelYzXcaOyhkBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"daf78cae93c244cf4d9273a6739b15042b65b6604ba76792faed59fe4df19285","last_reissued_at":"2026-07-05T10:47:02.590676Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:47:02.590676Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2504.05586","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":"2026-07-05T10:47:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"NkYbttEMuo+bPCeg7sljyN9ZhUtxoitdc5bxn548BY5NmLlcn992C4l68/IwJRqulDEZST9tOTkHcrsROnv9BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-21T01:06:26.388185Z"},"content_sha256":"ab440a4d6f088af483f65be49056c463530754858c69341400f985b9f2960efc","schema_version":"1.0","event_id":"sha256:ab440a4d6f088af483f65be49056c463530754858c69341400f985b9f2960efc"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:3L3YZLUTYJCM6TMSOOTHHGYVAQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Finding Fantastic Experts in MoEs: A Unified Study for Expert Dropping Strategies and Observations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ajay Jaiswal, Chong Wang, Jianyu Wang, Pingzhi Li, Ruoming Pang, Tianlong Chen, Xianzhi Du, Yixiao Li, Zhangyang Wang","submitted_at":"2025-04-08T00:49:08Z","abstract_excerpt":"Sparsely activated Mixture-of-Experts (SMoE) has shown promise in scaling up the learning capacity of neural networks. However, vanilla SMoEs have issues such as expert redundancy and heavy memory requirements, making them inefficient and non-scalable, especially for resource-constrained scenarios. Expert-level sparsification of SMoEs involves pruning the least important experts to address these limitations. In this work, we aim to address three questions: (1) What is the best recipe to identify the least knowledgeable subset of experts that can be dropped with minimal impact on performance? ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.05586","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/2504.05586/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":"2026-07-05T10:47:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"KVgw89+6t2/weQE9N+EO5ydKNMd7MXW5qIaNmJKDnKyqLruscJmIuLhqSZjiB/y8ksg9Gnr4wQJlDtQhmTNlCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-21T01:06:26.388561Z"},"content_sha256":"957579e41c3e54be16730949153477449985dda4b6e0b88d00a12102587d292b","schema_version":"1.0","event_id":"sha256:957579e41c3e54be16730949153477449985dda4b6e0b88d00a12102587d292b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ/bundle.json","state_url":"https://pith.science/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ/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":"2026-08-21T01:06:26Z","links":{"resolver":"https://pith.science/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ","bundle":"https://pith.science/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ/bundle.json","state":"https://pith.science/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3L3YZLUTYJCM6TMSOOTHHGYVAQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:3L3YZLUTYJCM6TMSOOTHHGYVAQ","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":"acfde7fe45a829cfbf3ee48a7bbde2cd258f462191cbf28439ea1af34ad88b35","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-08T00:49:08Z","title_canon_sha256":"cce9f0ef37c70e6efbafee2cbd4fd50c5e4e3485b5e25517f4f50f0424154b16"},"schema_version":"1.0","source":{"id":"2504.05586","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.05586","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"arxiv_version","alias_value":"2504.05586v2","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.05586","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"pith_short_12","alias_value":"3L3YZLUTYJCM","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"pith_short_16","alias_value":"3L3YZLUTYJCM6TMS","created_at":"2026-07-05T10:47:02Z"},{"alias_kind":"pith_short_8","alias_value":"3L3YZLUT","created_at":"2026-07-05T10:47:02Z"}],"graph_snapshots":[{"event_id":"sha256:957579e41c3e54be16730949153477449985dda4b6e0b88d00a12102587d292b","target":"graph","created_at":"2026-07-05T10:47:02Z","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/2504.05586/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Sparsely activated Mixture-of-Experts (SMoE) has shown promise in scaling up the learning capacity of neural networks. However, vanilla SMoEs have issues such as expert redundancy and heavy memory requirements, making them inefficient and non-scalable, especially for resource-constrained scenarios. Expert-level sparsification of SMoEs involves pruning the least important experts to address these limitations. In this work, we aim to address three questions: (1) What is the best recipe to identify the least knowledgeable subset of experts that can be dropped with minimal impact on performance? (","authors_text":"Ajay Jaiswal, Chong Wang, Jianyu Wang, Pingzhi Li, Ruoming Pang, Tianlong Chen, Xianzhi Du, Yixiao Li, Zhangyang Wang","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-08T00:49:08Z","title":"Finding Fantastic Experts in MoEs: A Unified Study for Expert Dropping Strategies and Observations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.05586","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:ab440a4d6f088af483f65be49056c463530754858c69341400f985b9f2960efc","target":"record","created_at":"2026-07-05T10:47:02Z","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":"acfde7fe45a829cfbf3ee48a7bbde2cd258f462191cbf28439ea1af34ad88b35","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-08T00:49:08Z","title_canon_sha256":"cce9f0ef37c70e6efbafee2cbd4fd50c5e4e3485b5e25517f4f50f0424154b16"},"schema_version":"1.0","source":{"id":"2504.05586","kind":"arxiv","version":2}},"canonical_sha256":"daf78cae93c244cf4d9273a6739b15042b65b6604ba76792faed59fe4df19285","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"daf78cae93c244cf4d9273a6739b15042b65b6604ba76792faed59fe4df19285","first_computed_at":"2026-07-05T10:47:02.590676Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:47:02.590676Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"kowCY/RP84fyjPB7rlUuVBjG3AImKxLpGMIh3FrakqAZFgA2g1B1eNwTCGQFcirWxPaP7bZhQelYzXcaOyhkBA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:47:02.591190Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.05586","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ab440a4d6f088af483f65be49056c463530754858c69341400f985b9f2960efc","sha256:957579e41c3e54be16730949153477449985dda4b6e0b88d00a12102587d292b"],"state_sha256":"e531917270e33ccafe73d5ffc7a5f6ae16729a9adf4d0e0e2b2a43a2d73e0e39"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4/YMO2c0cyAC5tf0i1rXaV1qMIaGcODp5AKZ+YykpPAp32ZUO4yR0I+dnp78Jxi3TB3Gy42HmLMQUNDIebNgAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-21T01:06:26.390819Z","bundle_sha256":"9ea688606c14cb65141951932c9d20b6b99d8d77b9bbd58de4100f637afd2764"}}